DayVectors

* 22 entries including: US Constitution (series), understanding AI (series), robocars & moral quandaries, Switzerland's Crypto Valley, chemical synthesis grand plan, USGOV versus Microsoft on data warrants, rebuilding US nuclear arsenal, blockchain and energy markets, trying to suppress robocalls, and Amazon getting into healthcare.

[FRI 30 MAR 18] NEWS COMMENTARY FOR MARCH 2018

* NEWS COMMENTARY FOR MARCH 2018: As discussed by an article from TIME.com ("Why China's Xi Jinping May Have to Rule for Life" by Charlie Campbell, 12 March 2018), on 11 March 2018, China's National People's Congress voted 2,958 to 2 to remove term limits on Chinese President Xi Jinping. The Chinese presidency is largely ceremonial, Xi having far more power as Chinese Communist Party (CCP) general secretary -- which doesn't have a term limit -- but the vote clearly showed that Xi had succeeded in placing himself at the center of all Chinese power.

The emergence of Xi as a strongman brings up unpleasant memories of the rule of the capricious Mao Zedong. Xi is no Mao, preferring tailored suits to People's Liberation Army field uniforms, and playing the role of international statesman instead of perpetual revolutionary. However, Xi's China is much more powerful on the world stage than Mao's China was, and the potential for global harm that much greater.

Professor Kerry Brown -- director of the Lau China Institute at King's College, London, and author of CEO, CHINA: THE RISE OF XI JINPING -- sees Xi's rise as taking place against an appalling passivity: "There was more opposition in Mao's time than today. It's kind of bizarre, because the consequences of opposing then were so high, but people still did it. And now you've got almost complete silence."

What is puzzling is why the 64-year-old Xi thought it necessary to centralize power unto himself. Deng Xiaoping, who more than any one person formed the modern Chinese state out of the chaos that Mao had left behind him, worked for consensus leadership, having learned from hard experience that concentration of power led to arbitrary and inept rule, as well as inevitable succession crises.

One reason for the turnabout appears to be that Xi's immediate predecessors -- Jiang Zemin and then Hu Jintao -- did encourage greater debate in party ranks, with result of increasing factionalism that threatened to destabilize the CCP. Xi has imposed order back onto the CCP through his anti-corruption campaign. He doesn't need to have his opponents executed; they are frightened enough of being publicly humiliated, and then thrown on the discard heap.

Xi has been given leash because the world has become a more fearful place for China. Chinese leadership hasn't forgotten the fall of the Soviet Union; the uprisings against autocratic regimes in the course of the Arab Spring reinforced that paranoia, as did the shift to the Right of many governments following the 2008 global financial crisis. While Chinese leadership appreciates that Donald Trump, the manifestation of that trend in the USA, is a noisy and inept buffoon, that in itself makes them nervous. As Brown puts it, Donald Trump has made people think: "Wow, this is a really unstable and unpredictable world where there might well be a trade war, and we need stable, predictable leadership in China to be able to deal with this."

Nonetheless, it appears that under the cloak of censorship, there is an undercurrent of public discomfort with Xi's heavy-handed leadership. Mao, though a frightening individual, was more revered than feared; Xi, though less frightening, is more feared than revered. Having centered all power on himself, he has reached a zenith, with the consequence that he has nowhere to go but down. Having purged and walked all over senior CCP officials, Xi has made a lot of enemies, and he can never relax his grip for fear of retribution.

Will there be a doddering old Xi still in power twenty years from now? Who wants that? A later generation of Chinese may well see Xi as a step backward. THE ECONOMIST commented that while Xi "is trying to boost China's image globally as a modern, outward-looking, and responsible state, the political system he governs seems pre-modern, opaque, and treacherous."

Following the vote eliminating presidential term limits on Xi, a Chinese blogger posted an image of Winnie the Pooh hugging a jar of honey, with the caption: "Find the thing you love and stick with it." Pooh Bear, it turns out, is seen by Chinese as a convenient stand-in for the comfortably-padded Xi. Censors quickly yanked the image. That by itself said much about the modern Chinese state: how neurotic and brittle a system, to fear the gentlest of digs via a children's tale. The digs, of course, will become ever more subtle and diverse until the censors read them into everything.

* In parallel news, on 18 March Vladimir Putin was elected president of Russia for a fourth six-year term. Ian Bremmer, writing on TIME.com ("Putin Won, But Russia Is Losing", 22 March 2018), found Putin's electoral victory distinctly hollow, saying that Putin is ...

BEGIN QUOTE:

END QUOTE

As Bremmer points out, Russia's economy is smaller than that of Canada; despite Putin's aggressive swagger, Europe's military establishment handily outmatches that of Russia; and Russia's allies are a loser's club, members including Venezuela, Cuba, Sudan, North Korea, Syria, and Serbia. China makes deals with Russia -- but on Chinese terms.

Putin is an opportunist, seeking a short-term advantage while blind to long-term liabilities. The Russian seizure of Crimea in 2014 pumped up his tough-guy image -- but the loss of the most pro-Russian component of Ukraine meant the bolstering of anti-Russian sentiment in the remainder, with Moscow in control of territory that was more an economic drain than an asset. The Russians already had a naval port to the Black Sea, while Western sanctions contributed to a drop in Russian GDP from 2014 to 2015 estimated at 35%.

The Baltic states are now in firmly in the Western camp; NATO troops are even stationed there, in response to Russian hostility. NATO didn't want to put forces into the Baltics, but Putin forced NATO's hand -- and by the same coin, boosted a European defense buildup. Azerbaijan and the Central Asian states are more interested in long-term ties with growing China than with rusty Russia. In the regional influence game, Beijing's shadow is falling over the Kremlin. As discussed here last month, Putin's involvement in Syria has bought him little but trouble.

Putin's most staggering blunder was to direct his intelligence services to attack the West, proving only too successful in the US elections of 2016. He helped deny the presidency to Hillary Clinton, seen as an enemy of Russia, with the election of Trump also throwing America into confusion and contention. Both these gains were psychological, not practical: the US is at least as hostile to Russia as it would have been had Clinton been elected, with a backlash still growing in force against internet-based subversion. The Russians, in leveraging off the malign efforts of home-grown malcontents, have made them into targets. Now Russian intelligence has been accused of poisoning Sergei Skripal, a former double agent exiled in the UK, leading to aligned condemnation in the West. Putin's bland denials of involvement are greeted with acid laughter.

Putin's primary goal in his cyber-attacks on the West was to raise the stature of his own regime by making the governments of adversaries look worse; he hoped to pump up his image, while distracting the Russian public from corruption and economic stagnation at home. Russia's economic position is improving for the moment, but the country still remains dependent on oil production. Putin's distaste for the USA can only be reinforced by the way the American fracking boom has helped hold down oil prices. Once the shift away from a carbon-based economy begins in earnest -- as it surely will sooner or later -- Russia's oil-based economy will be faced with persistent decline, and ultimately collapse.

Bremmer concludes that Putin's attempts at economic diversification have been ineffectual, and leave him at a dead end:

BEGIN QUOTE:

END QUOTE

ED: The 20-teens saw a global shift towards authoritarianism; the authoritarian regimes demonstrating they have little going for them, might the 2020s see the pendulum swing back towards liberal democracy again? As Winston Churchill put it, democracy is the worst system there is, except for all the others -- which is exactly what we've been finding out.

Certainly, it's a very good bet that Donald Trump will be given the boot out of the White House in 2020. He's the least effective and most comical of the authoritarians, but the end of his misrule should suggest to his counterparts elsewhere that their own misrule is living on borrowed time. After we see the back of Trump, we may hope see the back of Putin, the Sawdust Stalin.

* In early February, the US Supreme Court upheld a Pennsylvania state court decision that the state's electoral map had been "gerrymandered" by the state Republican apparatus, with electoral districts redrawn to guarantee they would stay Republican. The gerrymandering was spectacularly obvious in the case of the infamous 7th district, which was so comically tortured in appearance that it was nicknamed "Goofy kicking Donald Duck". There are other gerrymandering cases before the court, but as discussed by an article from BLOOMBERG.com ("The Supreme Court Is Finally Tackling Gerrymandering" by Peter Coy & Greg Stohr, 19 January 2018), it appears that, thanks to the courts, the long era of gerrymandering in the USA is coming to a close.

The Supreme Court traditionally has let gerrymandering alone, since it inevitably lends a politically partisan tone to the court. Political scientists, statisticians, and other number-crunchers have made a case to reassure the justices that it actually is possible to impartially judge on gerrymandering cases. After all, gerrymandered districts are set up using analytical tools to find a partisan advantage; effectively the same tools can be used to ensure a level playing field. "We totally know how to do this," says Gary King, a political scientist at Harvard, who's consulted for both major parties as well as independents and courts on redistricting issues.

King and a colleague, Robert Browning, came up with the principle of "partisan symmetry" in 1987. It says, simply enough, that districts should be drawn so the parties would achieve the same outcomes given the same number of votes. For example, if the lines are drawn such that Republicans win 80% of a state's seats and only have 55% of the votes, that's fine -- as long as Democrats would also win 80% of the seats when they have 55% of the votes.

King and two colleagues offered the partisan symmetry standard to the Supreme Court in a 2006 redistricting case, LULAC V. PERRY. Some of the SCOTUS justices were sympathetic to the argument. There are other suggestions for assessing partisanship of districting, but King warns that districting should be defined by broad principles and not detailed criteria. The detailed approach would encourage partisans to game the system, "crawling right up to the edge"; given partisan symmetry, nobody will be inclined to rock the boat, knowing that they would lose a court challenge.

As usual, the swing vote could be Justice Anthony Kennedy. In a 2004 case, he voted with the majority in upholding a Pennsylvania congressional map, but he did not rule out the potential for finding a standard. In any case, it seems like the SCOTUS is finally getting serious about tackling gerrymandering -- if, to a degree, because it's a damned nuisance that they wish would just go away.

ED: One of the ongoing cases before SCOTUS on gerrymandering is a Republican complaint over Democratic gerrymandering. Yes, the Democrats are hardly blameless in the gerrymandering game; and taking them to task, rightly or wrongly, is all for the good. The two parties would be best off agreeing on some computer algorithm, devised by an impartial research group, to do the districting. That way, any deviation from the algorithm could be easily checked, and then challenged in court, where it would be immediately shot down. In any case, it finally seems that gerrymandering is on the way out.

[THU 29 MAR 18] WINGS & WEAPONS

* WINGS & WEAPONS: The US Army's "MGM-140 Army Tactical Missile System (ATACMS)" AKA "Attack 'Ems" -- discussed here in 2017 -- was developed in the 1980s, and is now out of production. There's plenty of inventory for the time being, but as discussed by an article from AVIATIONWEEK.com, ("Raytheon To Flight Test DeepStrike Missile In 2019" by James Drew, 10 October 2017), the Army is now searching for a successor under the "Long-Range Precision Fires (LRPF)" program.

Raytheon and Lockheed Martin were awarded contracts in May 2017 to develop competing prototypes under a 34-month technology risk-reduction phase. Raytheon display a mockup of the company's "DeepStrike" missile at an October exposition in Washington DC, along with mockups of Raytheon's proposed land-based Standard Missile-3 and SkyCeptor interceptors.

The goal of the LRPF program is to field a smaller, longer-range ATACMS successor. The minimum objective range for the follow-on LRPF is 400 kilometers (250 miles), 100 kilometers more than ATACMs. The LPRF cannot have a range of more than 500 kilometers (310 miles) because of the Intermediate-Range Nuclear Forces Treaty with Russia, which sets that as the limit for tactical ballistic missiles.

Raytheon is moving quickly, the intent being to have a prototype flying by 2019. The tactical launcher will be Lockheed's "M142 High Mobility Artillery Rocket System (HIMARS)", a wheeled vehicle that can haul one ATACMS or six Guided Multiple Launch Rocket Systems (GMLRS) rockets. The Army wants to HIMARS to haul two LRPF missiles. That means Orbital ATK -- the effective sole supplier of solid-fuel rockets to both Raytheon and Lockheed Martin -- will need to come up with a more compact propulsion system. The LRPF will use GPS guidance; it isn't clear what kind of targeting seeker it will carry, and if it will have antiship capability.

As a footnote, in a demonstration on 22 October 2017, a HIMARS launcher carried on the USS ANCHORAGE, a landing transport, unloaded a GMLRS rocket on a ground target 70 kilometers (43.5 miles) away. The ANCHORAGE is normally only fitted with defensive armament; the ability to perform strikes with GMLRS rockets or ATACMS missiles grants it a bit of offensive punch. It's not really a new idea. During World War II, Allied landing ships were sometimes rigged to fire arrays of unguided rockets to smash enemy beach defenses. HIMARS munitions, however, have much greater accuracy and range.

* As discussed by an article from WIRED.com ("Boeing Planes Could Fire Lasers From Their Noses to Spot Turbulence" by Eric Adams, 15 September 2017), turbulence is a problem for airlines. According to the US Federal Aviation Administration, 44 people were severely injured by turbulence in 2016, and that doesn't count the less severe roughing-up and spilled drinks passengers suffer on flights normally.

To deal with the problem, aerospace giant Boeing is fitting a 777 jetliner with a lidar -- light radar -- system in the nose to spot turbulence on the flight path ahead. The system projects a laser beam ahead of the aircraft, while an optical sensor tracks light reflected back by dust particles along the path of the beam. The wavelengths of radars are too long to pick up dust; radars can handle water droplets, but that doesn't help much in clear sky.

Software analyzes the aircraft's velocity relative to the movement and velocities of particles at different distances. Significant changes in the velocity differentials -- like pockets of air moving faster than the stuff around it -- reveal turbulence ahead, with the system then alerting the flight crew with audible and visual cues integrated into the cockpit system. It can only give about a minute's warning, but that should be enough to tell everyone on board to sit down and hang on. At the present time, all the aircrew has available to know about turbulence is reports from other aircraft, and a general understanding of weather conditions along the flight path.

The lidar system was originally developed by the Japan Aerospace Exploration Agency (JAXA), which has been collaborating with Boeing since 2010 to configure it for use on commercial aircraft. There are ground-based systems in use right now, but they're much too heavy for aircraft. Boeing's test system weighs about 84 kilos (185 pounds), and only draws about 3.3 watts of power.

The exercise is part of Boeing's "ecoDemonstrator" program. About every year and a half, the company selects a set of emerging technologies, then flies them on a test aircraft for six weeks.

* The guided 12.7-millimeter (0.50 caliber) developed by Orbital ATK for the US Defense Advanced Research Projects Agency (DARPA) under the "Extreme Accuracy Tasked Ordnance (EXACTO)" program was discussed here in 2015. Now Orbital ATK is developing a 30 x 173-millimeter guided round, for use in the 30-millimeter Chain Gun and other automatic cannon, obviously using similar technology.

The guided round will be directed by a laser-based fire-control system, which will track the round and give it course corrections, via the laser and an optical sensor on the round. The round will then activate actuators to change direction -- all the smarts, in other words, are in the fire-control system. Orbital ATK is also working on proximity-fuzed and airburst 30 x 173 rounds.

Incidentally, on poking around online on this topic, I ran across mention of a sniper rifle using 30-millimeter rounds. While 12.7-millimeter "anti-material rifles" are commonplace, and it appears 20-millimeter rifles not unknown, it is hard to know if 30-millimeter rifles are for real. There are certainly fictional references to them, and possibly some one-off designs for them -- but how a kick that would be like a sledgehammer would be managed is an interesting question. Nonetheless, it would certainly be impressive!

[WED 28 MAR 18] OBEY THE LAW

* OBEY THE LAW: As discussed by an article from WIRED.com ("Lawyers, Not Ethicists, Will Solve the Robocar Trolley Problem" by Aarian Marshall, 28 May 2017), now that we're headed for the day of the robot car, moral questions are coming out of the woodwork.

Consider, for example, the scenario of the "trolley problem". Supposed Bob sees a runaway trolley barreling down the tracks toward five people. Bob is by railway switch that allows him to divert the trolley to another track, where just one person stands. Does he decide to spare five lives at the expense of one? In a sense, we can ask the same question of a robocar; does it spare the lives of five pedestrians at the expense of the life of a passenger?

Of course, the robocar doesn't make that decision; the decision is inherent in the design of its software, and the question is one of software design. In a common-sensible paper published in NORTHWESTERN UNIVERSITY LAW REVIEW, Stanford University researcher Bryan Casey dismissed the trolley question, saying that software engineers will be "less concerned with esoteric questions of right and wrong than with concrete questions of predictive legal liability."

In other words, companies building robocars are not going to pay much attention to academic moral arguments; they will instead be getting their directives from their legal staff and the courts, as well as from the desires of customers. Casey writes: "The trolley problem presents already solved issues -- and we solve them democratically through a combination of legal liability and consumer psychology."

Casey points to Tesla as an example. Drivers of the Tesla EVs can put the vehicle on Autopilot, and let the machine cruise. Rather than guess at what speed would be best, Tesla's engineers decided it would be best just to follow the speed limit, on the assumption that the authorities have thought out what the best speed is. Following the speed limit also gets Tesla off the hook in the case of an accident. Writes Casey:

BEGIN QUOTE:

END QUOTE

Of course, there's also the issue of what customers want. In January, Tesla sent out an Autopilot update that lets cars zip along at up to 5 MPH over the limit on some roads, after owners complained about getting passed by everyone else. If that's what customers want, that's what they get -- at least, if it's not demonstrably improper to give it to them. As Tesla officials understand, just because customers ask to be allowed to do something hazardous doesn't mean Tesla can dodge the liability bullet if things go dangerously wrong.

Noah Goodall -- a transportation researcher with the Virginia Transportation Research Council who studies self-driving cars -- doesn't see that the trolley problem brings much to the discussion of robocar safety: "Trolley problems are pretty unrealistic -- they throw people."

Goodall adds that getting customers to feel comfortable when riding in a robocar means "being upfront about what these vehicles really do. They prevent a lot of crashes, a lot of deaths. Fine-tuning things are difficult, but companies should prove they put some thought into it."

More than 35,000 people die on American roads every year; over 1.25 million people die around the globe, a carnage on the level of that of a world war. That's the real moral bottom line, and the trolley question vanishes next to it. Nobody is going to put any serious effort into designing machines to make moral decisions that, in practice, stymie humans.

[TUE 27 MAR 18] CRYPTOSWISSTEMS

* CRYPTOSWISSTEMS: As discussed by an article from ECONOMIST.com ("Tales From The Crypto-Nation", 24 February 2018), the town of Zug in Switzerland is not well-known to anyone outside the country -- except to the global cryptological community, which calls it "Crypto Valley", the place being a hotspot for firms working on cryptosystems.

Zug's cryptological history traces back at least to World War II, when a Swede named Boris Hagelin came to Zug to set up an operation to build his ingenious cipher machine. Being a little mechanical marvel, the Hagelin cipher machine was a natural fit for the Swiss, with a tradition of watchmaking and other fine mechanical products.

Cryptology later moved on to digital systems, with Zug keeping pace. Now the town is buzzing over cryptocurrencies -- a fit with financial systems, another Swiss specialty -- as well as commercial and government security systems. Johann Schneider-Ammann, Switzerland's economy minister, says that the country should become the "crypto-nation" -- of course, with the Crypto Valley at the center of the action.

To that end, Switzerland is loosening the regulation of cryptosystems, even as regulations are being tightened up in other countries. Swiss firms are coming up with schemes to archive cryptographic keys in cold, dry, secret sites, protected by response teams. What better place than an old military bunker in the Alps? Even as cryptocurrencies create turmoil around the world, Zug remains friendly to them, with "Bitcoin Accepted Here" signs in evidence around town -- at least for now. Less controversially, residents can get a blockchain-based digital identity.

About a quarter of the world's total of $5 billion USD in initial coin offerings (ICO) -- a form of crowdfunding, in which investors are issued digital tokens giving a share in a project -- was raised in Switzerland. Of the ten biggest ICOs, four in part came out of Zug. The town's laws are friendly to crypto entrepreneurs, and taxes are low. After World War II, Zug cut its corporate tax rate to 8.5%; today it's 14.6%, still low compared to Zurich's 21%.

Crypto valley went into high gear in 2013 when the Etherum Foundation, a non-profit working on a blockchain system known as "Etherum", settled in the town. Now there's over 150 firms there. Blockchain companies have been on a particular boom, with Zug Mayor Dolfi Mueller saying: "They landed like flying saucers."

Local tax authorities, accountants, and lawyers are all conversant in blockchain tech. Switzerland's light-touch central government, tradition of direct democracy, and inclination to libertarian views underpin Crypto Valley's success, by cultivating an environment that gives crypto entrepreneurs space for action. Other potential hubs like Hong Kong and Singapore aren't as enticing.

Switzerland's long reputation as a safe banking haven also supports the crypto boom. According to Niklas Nikolajsen of Bitcoin Suisse, a financial-services provider: "You can have all the armored walls in the world, but if your vault is in China or Singapore and the government says: 'I'm seizing your assets!' -- there's nothing you can do. That would never happen in Switzerland."

Joerg Gasser, the state secretary for international finance, has little doubt that if -- when -- the bitcoin bubble bursts, investors will demand regulation, but the crypto sector will not have cause to fear being regulated to death. At the same time, Gasser is not about to tolerate bandits on his watch. The national regulator, FINMA, is investigating several ICOs for possible breaches of regulations, such as anti-money laundering rules. A working group is now examining the rules system for ICOs -- the idea not being to become hostile to crypto entrepreneurs, but to create an orderly environment where they can flourish.

Can Crypto Valley endure? Cryptotech is based largely on brainpower; the material resources needed to support the exercise can be found almost anywhere in the developed world. However, there's something to be said for a place where those interested can come together to trade knowledge, to compete, and to grow.

[MON 26 MAR 18] UNDERSTANDING AI (17)

* UNDERSTANDING AI (17): The legal issues with pervasive, AI-driven surveillance are going to be worked out, though it may take a generation. As for AI startups, their immediate concern is making sure their technology works as advertised, and doesn't get them into any unnecessary trouble. Much of that trouble promises to come from taking a new and unproven technology too much for granted.

Alex Hauptmann, a professor at Carnegie Mellon University who studies the social implications of technology, points out that computers are faced with a number of fundamental challenges in analyzing video. One of the problems is resolution. Consider a neural network that's been trained to analyze human actions in a video. It does so by reducing human bodies to "stick figures" -- assemblies of body elements, such as arms, legs, shoulders, heads, and so on -- and then observing how the stick figures change from one video frame to the next. The AI can then determine, for example, if someone's running, or brushing their hair. However, as Hauptmann points out:

BEGIN QUOTE:

END QUOTE

As anybody who sees security camera footage on TV realizes, the imagery is often grainy, and the camera angles are weird. Hauptmann offers as an example a security camera in a convenience store that is targeted on the cash register, but also takes in the window facing the street. If a mugging takes place outside, where the camera can't get a good view of it, a human watching the video stream might be able to figure out what's happening, but the AI wouldn't have a clue. Similarly, even if AI can see someone running, it can't figure yet figure out the context. Is the person running in apprehension from a crime? Or just trying to catch a bus?

There are things AIs can do very well today, within bounds. Hauptmann says license plate tracking to follow vehicles is "a solved problem for practical purposes," and facial recognition in controlled settings is also solid -- though facial recognition in poor-resolution CCTV video is much more troublesome. Identifying things like cars and items of clothing also works well; automatically tracking one person across multiple cameras can be done, but only if conditions are good. Hauptmann says: "You're pretty good at tracking an individual in a non-crowded scene -- but in a crowded scene, forget it." He says it's very tough if the individual is wearing nondescript clothing.

Privacy activists and other critics have also suggested that AI systems need to be designed to be impartial, and not trained to reflect traditional prejudices -- for example, giving excessive and occasionally unjust scrutiny to black people, while being inclined to give white people a free pass. More significantly, what happens if an AI misinterprets a situation, calls it in as an emergency, and an innocent bystander gets killed as a result?

Okay, so our AI systems still leave much to be desired. On the nasty side of that coin, they're still capable enough to be frightening -- as China has demonstrated in its western Xinjiang province, where the state is using high tech to clamp down on its restless Muslim Uighur ethnic group. In Xinjiang, traditional methods of surveillance and civil control are being combined with facial recognition, license plate scanners, iris scanners, ubiquitous CCTV, and a DNA database to create a "total surveillance state", where individuals in public spaces are tracked constantly.

Xinjiang has been described as "a laboratory for high-tech social controls" by THE WALL STREET JOURNAL. China is working towards a "police cloud" network of security cameras, and police are now increasingly wearing video glasses with facial-recognition capabilities. The Chinese are becoming masters of 21st-century surveillance technology, and are selling it on the export market.

A similar infrastructure is being put together in Moscow, likely with Chinese help -- facial recognition software being plugged into a centralized system of more than 100,000 high-resolution cameras that cover more than 90% of the city's apartment entrances. The widespread use of AI in China, and increasingly in Russia, means the technology improves on itself as it gets more training in widespread practical service.

Nobody is so worried about the emergence of such a malevolent surveillance state in the West -- but as pointed out earlier, there's nothing to stop police from operating fleets of patrol drones, so we're headed for a surveillance society of our own. While there's been some discussion about what restraints need to be placed on such an all-seeing eye, we're a long way from having established rules to keep the system from playing Big Brother.

The AI startups will have inputs into establishing the rules, but obviously they can't be the ones making those rules; it's a matter that will have to be considered by society as a whole. Of course, none of them believe that AI is an inherently flawed and dangerous technology that needs to be banned. Matt Sailor feels the AI systems have more potential for good than harm: "With any new technology there's a danger it could fall into the wrong hands. That's true of any technology ... and I think the pros in this aspect greatly outweigh the cons." [TO BE CONTINUED]

[FRI 23 MAR 18] AMERICA'S CONSTITUTION (7)

* AMERICA'S CONSTITUTION (7): While the new Congress clearly had its roots in the Confederation Congress, it was something different. The Continental Congress had come together as an ad-hoc organization, operating on an improvised and shaky legal basis. Even the Articles of Confederation hadn't granted the following Confederation Congress any "legislative powers" to make laws.

As or more significantly all authority, such as there was, under the Articles of Confederation had resided in the Confederation Congress. The new Congress, in contrast, was only one of three arms of government, established as the legislative, executive, and judicial branches. The legislative branch, the Congress, was also split into two different assemblies, the Senate and the House, with different structures, rights, and responsibilities.

They had to cooperate to get things done, with both assemblies having to approve a bill for it to be passed. Each could devise and pass its own bills -- the House had an exclusive right to come up with bills raising revenue, but it was a meaningless distinction, the Senate could informally conceive a bill, with the House deciding whether it wanted to formally back it. However, for the bill to mean anything, the two assemblies had to come up with a common bill in conference. A president could veto a bill, in which case Congress could only override it with a two-thirds majority in both chamber. If the judiciary decided a bill was unconstitutional, it was dead, and there was nothing Congress could do about it -- other than try again, revising the bill if possible to meet judicial objections.

Each state got two senators in the senate, each senator having one vote. Each state carried the same weight in the senate, regardless of a state's size; of course, that was a measure to reassure the smaller states that they couldn't be easily pushed around by the bigger ones. Senators, as with the Confederation Congress, were to be selected by state legislatures -- a reality generally forgotten today. As stated in Section 4, exactly how that was to be done was left up to the states, though Congress reserved a constrained right to intervene.

Unlike the Confederation Congressmen, however, senators served six-year terms, not one-year terms; they also could not be arbitrarily recalled from the Senate by a state legislature. At the outset, the terms were staggered by two years, so an effective third of the Senate would be up for election every two years.

The representatives in the lower House were elected from districts set up in a state, not selected by state legislatures, with the number of districts proportional to the population of the state. The Constitution was very loose in defining how the states set up districts for representatives, saying only that there had to be at least one for a state, and not more than one for each 30,000 inhabitants of the state. Effectively, the Framers passed the buck on the issue, on the basis that Congress would establish laws to clarify districting.

In any case, the House was to be based on proportional representation, with larger states having more clout there than smaller ones. Again, each representative got a single vote. All representatives served a two-year term, with the entire House being up for election every two years.

The regularly-scheduled elections were a rejection of the much more irregular British system of, effectively, simply "holding elections when necessary, or every now and then". Retaining the British system presented a threat of an existing Congress clinging to power, refusing to be displaced by a new Congress; or an existing Congress selectively calling elections only when it was in their interests to do so. Regular scheduling of elections also denied a president the ability to dissolve Congress and call elections. Incidentally, such electoral irregularity still persists in Britain. There is clearly plenty of method to the British electoral madness; but Americans, used to predictable elections, have trouble figuring out what it is.

Both senators and representatives were paid by the Federal government, not by the states as had been the case in the Confederation Congress; and there was no requirement that the senators or representatives of a state vote as a bloc, or be members of the same party. There were no term limits on the members of Congress; they could remain in their seats for as long as the state legislatures (for senators) or voters (for representatives) wanted to keep them there. However, unlike the old Congress, the new Congress could censure or even evict its own members.

The bicameral organization of Congress was not just a question of balancing the influence of the states, giving them all equal voices in the Senate proportional representation in the House. That was a neat compromise, to be sure, but the bicameral structure also ensured that the two elements of Congress would act as a check on each other: if the powerful senators reached too far, the House could restrain them; while if the populist representatives pandered to the mob, the Senate could restrain them in turn. The Senate and House had to cooperate to pass laws -- with the back-&-forth involved helping to ensure, if by no means guaranteeing, that laws were carefully defined.

There's a popular notion that this division of powers was implemented to limit the generation of laws by Congress, but there's not much evidence to support that idea. The scheme wasn't designed to be efficient; that was exactly the point, to ensure that the legislature didn't become a single-minded bully. The fact that all members of Congress were sworn to uphold the Constitution gave them principles, most notably stated in the Preamble, to work from, and to be held accountable to if they disregarded them. [TO BE CONTINUED]

[THU 22 MAR 18] SPACE NEWS

* Space launches for February included:

-- 01 FEB 18 / KANOPUS-V 3 & 4, SMALLSATS -- A Soyuz 2-1a booster was launched from Vostochny at 0217 UTC (local time - 9) to put "Kanopus-V 3" and "Kanopus-V 4" Earth observation satellites into orbit for Roscomos. The two spacecraft, which each weighed 473 kilograms (1,043 pounds), and carried three color & grayscale imagers. They were to assist the Russian government in disaster response, mapping, forest fire detection, and resource monitoring.

The launch also included nine nanosatellites:

• S-Net x 4: Four 8.5-kilogram (18.7-pound) satellites from the Technical University of Berlin, to demonstrate an S-band inter-satellite communications package.
  
  
• Lemur 2 x 4: 3-unit (3U) CubeSats from US-based Spire Global, to track atmospheric conditions through GPS signal occultation. This was the third launch of Lemur 2 satellites in three weeks, following an Indian Polar Satellite Launch Vehicle flight and Rocket Lab's Electron rocket test flight in January.
  
  
• D-Star One Phoenix: An AMSAT (amateur radio satellite) in the 3U CubeSat configuration, developed by German Orbital Systems in Berlin in cooperation with the Czech company iSky Technology, was also launched Thursday. It replaced the D-Star One nanosatellite lost on a Soyuz launch in November.

-- 02 FEB 18 / CSES, SMALLSATS -- A Chinese Long March 2D booster was launched from Jiuquan at 0751 UTC (local time - 8) to put the "China Seismo-Electromagnetic Satellite (CSES)" AKA "Zhangheng 1", and six secondary payloads into orbit. Zhangheng 1 was a Chinese-led mission to study how electromagnetic signals in Earth's atmosphere and ionosphere could help predict earthquakes.

The satellite had a launch mass of 730 kilograms (1,600 pounds) and carried a suite of nine instruments, supported by six deployable booms. The mission was scheduled to last for five years. The satellite was operated by China National Space Administration (CNSA), along with the China Earthquake Administration and the China National Space Administration, in cooperation with the Italian Space Agency (ASI) Zhang Heng was a Chinese scholar who did some investigation of earthquakes. The secondary payloads included:

• GomX 4A, GomX 4B: A pair of 6U CubeSats, built by GomSpace of Denmark. GomX 4A was to obtain data for the Danish Ministry of Defense to track aircraft and ship movements over the Arctic region, while GomX 4B was an experimental nanosatellite funded by the European Space Agency. The twin satellites flew in long-distance formation at a range of up to 2,800 miles (4,500 kilometers) to test an S-band radio link between the craft. GomX 4B did the maneuvering, using a set of cold-gas thrusters; it also carried a hyperspectral multi-color camera, a star tracker, and components to be tested against space radiation. The GomX 4 satellites built on experience obtained on a predecessor nanosatellite named "GomX 3", released from the International Space Station in 2015.
  
  
• NuSat 4, NuSat 5: Earth-imaging microsatellites for Satellogic of Argentina. Named "Ada" and "Maryam", the satellites were built in Montevideo, Uruguay, with each having a launch mass of about 37 kilograms (80 pounds), and carrying a hyperspectral imager. They joined three earlier NuSat space platforms in orbit, these three having been launched by Chinese boosters in 2016 and 2017. Satellogic plans to launch dozens, possibly hundreds, of NuSats.
  
  
• Two 3U CubeSats, named "Fengmaniu (FMN) 1" and "Shaonian Xing", flown by Chinese start-up companies. FMN-1 was developed by Link Space Aerospace Technology for scientific education and technology demonstration; it carried two cameras, and also functioned as an AMSAT relay. Shaonian Xing AKA "Youth Star" was a student engagement exercise.

-- 03 FEB 18 / TRICOM 1R -- A JAXA SS-520-5 booster -- a modified sounding rocket with an added third stage -- was launched from Uchinoura at 0503 UTC (local time - 9) to put the TRICOM 1R satellite into orbit. TRICOM 1 was a 3U CubeSat with a store-&-forward communications system and an imaging camera. This was the smallest booster to ever successfully put a satellite into orbit. It was a reflight, following a launch failure in January 2017. The Uchinoura site, previously known as Kagoshima, is used for launching sounding rockets.

-- 06 FEB 18 / FALCON HEAVY TRIAL FLIGHT -- The first SpaceX Falcon 9 Heavy booster was launched from Cape Canaveral at 2045 UTC (local time + 5) on a trials flight. The heavy-lift rocket was made up of three Falcon 9 rocket cores strapped together, with 27 Merlin 1D engines burning at liftoff. The strapon boosters performed a simultaneous soft landing at Cape Canaveral; the core stage failed to land on the recovery barge. The booster carried a dummy payload, in the form of Elon Musk's Tesla roadster, with a dummy at the wheel; the roadster was sent off on a trajectory towards the asteroid belt.

-- 12 FEB 18 / BEIDOU x 2 -- A Chinese Long March 3B booster was launched from Xichang at 0510 UTC (local time - 8) to put the "Beidou 3 M3" and "Beidou 3 M4" navigation satellites into orbit. They were placed in a medium Earth orbit with an altitude of 13,350 miles (21,500 kilometers) and an inclination of 55 degrees. They were third-generation satellites, the 28th and 29th Beidou satellites to be launched.

-- 13 FEB 18 / PROGRESS 69P (ISS) -- A Soyuz booster was launched from Baikonur at 0813 UTC (local time - 6) to put the "Progress 59P" AKA "MS-08" tanker-freighter spacecraft into orbit on an International Space Station (ISS) supply mission. The capsule docked with the ISS two days after launch. It was the 69th Progress mission to the ISS.

-- 22 FEB 18 / PAZ, MICROSAT 2A & 2B -- A SpaceX Falcon 9 booster was launched from Vandenberg AFB at 1417 UTC (local time + 8) to put the "Paz" radar imaging satellite for Hisdesat of Madrid, Spain, which oversees Spain's governmental satellite programs.

Built by Airbus Defense and Space, Paz had a launch mass of 1,450 kilograms (3,200 pounds) and carried an X-band active-array synthetic aperture radar (SAR), with narrow-angle and wide-angle observation modes, best resolution being about 25 centimeters (10 inches). Paz was to obtain radar imagery of the Earth for military and commercial use. It also carried an AIS ship tracking system, working in conjunction with the SAR, and a GPS occultation receiver to monitor weather.

Paz shared its orbit with two German radar satellites -- TerraSAR X and TanDEM X -- launched in 2007 and 2010. They provided imaging of the Earth on a three-day revisit time, maximum, more typically one day. Imagery is available to all, though high-resolution imagery is only released to qualified clients.

The flight also carried two low-orbit comsats, named "Microsat 2a" and "Microsat 2b" AKA "Tintin A & B". They were prototypes for the SpaceX "Starlink" constellation, the idea being to put a broadband "internet in the sky" into orbit. According to SpaceX, each satellite was about the size of a mini-refrigerator and carried computer, power, command & control, propulsion, and GPS navigation gear typical for small spacecraft. A Ku-band broadband phased array antenna and an inter-satellite optical communications link were also carried on each spacecraft.

In an application seeking FCC approval of the broadband fleet, SpaceX outlined a network that could include 4,425 satellites operating in Ku-band and Ka-band frequencies. The Ku-band and Ka-band satellites would orbit at an altitude of from 1,110 kilometers to 1,325 kilometers (689 miles and 823 miles). Besides the huge constellation of Ku-band and Ka-band satellites, SpaceX also proposed deploying 7,518 spacecraft transmitting broadband signals in V-band frequencies from low-altitude orbits around 340 kilometers (210 miles) above the planet.

The Falcon 9 first stage had been flown on a previous mission in August 2017; it was not recovered on this flight. There was an attempt to recover the payload fairing, using a boat with a net, but it missed by a few hundred meters. Changes are to be made with the parachute system of the fairing to improve matters.

-- 27 FEB 18 / IGS OPTICAL 6 -- An H-2A booster was launched from Tanegashima at 0434 UTC (local time - 9) to put an "Information Gathering Satellite (IGS) Optical 6" military surveillance satellite into orbit. IGS Optical 6 was the 15th IGS satellite, from initial launch in 2003:

• The first two IGS satellites -- Optical 1 and Radar 1 -- were successfully launched together in March 2003 by a single H-IIA booster.
  
  
• A duplicate launch eight months later was a failure when one of the H-IIA's solid rocket boosters didn't separate, with the launch vehicle destroyed by range safety.
  
  
• A replacement optical satellite, IGS Optical 2, was launched in 2006.
  
  
• IGS Radar 2 was launched in February 2007, sharing the journey to orbit with IGS Optical 3V, which was a prototype for a second generation of optical imaging missions.
  
  
• The second-generation optical component of IGS consisted of IGS Optical 3 and 4, which were launched in November 2009 and September 2011 respectively.
  
  
• Radar satellites, IGS Radar 3 and 4, were deployed in December 2011 and January 2013, with the Radar 4 launch also including a prototype third-generation optical satellite, "IGS Optical 5V".
  
  
• An on-orbit spare radar satellite was launched in February 2015.
  
  
• The third-generation IGS Optical 5 was launched in March 2015, with IGS Radar 5 launched in March 2017. The radar satellites have not yet been updated to a third generation.

With the launch of IGS Optical 6, the constellation consisted of Optical 4, 5, & 6, plus Radar 2, 3, 4, & 5. All the IGS satellites were built by Mitsubishi Heavy Industries.

[WED 21 MAR 18] CHEMISTRY MOONSHOT

* CHEMISTRY MOONSHOT: An article from SCIENCEMAG.org ("Billion-Dollar Project Would Synthesize Hundreds Of Thousands Of Molecules In Search Of New Medicines" by Robert F. Service, 19 April 2017) focused on Martin Burke, a chemist at the University of Illinois in Urbana, who is proposing a chemistry "moonshot": an initiative to synthesize the majority of the hundred-thousands of known organic natural products: the diverse small molecules made by microbes, plants, and animals. Burke estimates that the effort would take 20 years and require a billion USD in funding.

Organic natural products make up more than half of all medicines, as well as dyes, diagnostic probes, perfumes, sweeteners, lotions, and so on. However, discovering, isolating, and testing new natural products is slow, painstaking work. For example, consider bryostatins -- a family of 20 natural products first isolated in 1976 from spongelike marine creatures known as bryozoans. Bryostatins have demonstrated potential for treating Alzheimer's disease and HIV, and demand has skyrocketed. Unfortunately, tons of bryozoans have to be mashed up to get a few grams of bryostatins, and synthesizing them has proven troublesome.

Burke thinks he can make synthesis much easier. In 2015, he and his colleagues unveiled a machine that can link a variety of building blocks, like Lego blocks, to create thousands of natural product compounds and their structural relatives. The machine was just a lab prototype, not a practical product, but Burke sees no problem with scaling it up. Molecular biologists have already automated the synthesis of short strands of DNA, proteins, and sugar chains, leading to a revolution in biomedicine. Burke says that doing the same for natural products "could have major positive implications not just for chemistry, but for society."

When Burke revealed his machine in 2015, he estimated that assembling 75% of natural products with his machine would take some 5,000 different building blocks, compared with just four for DNA. Synthesizing and stockpiling the building blocks would be logistically troublesome, calling the practicality of the scheme into question. However, Burke's lab recently teamed up with that of Jeffrey Skolnick, a computational biologist at the Georgia Institute of Technology in Atlanta, to seek simplification.

They surveyed the literature on natural products, counted 282,487 compounds, and mapped all their structures. Skolnick's team then designed an algorithm to break each compound into fragments, cutting only single bonds between carbon atoms -- the kind of bonds Burke's machine can reassemble. Armed with that data, the software determined how many unique fragments it would take to reconstruct the library. The answer was that 1,400 building blocks would be enough to synthesize 75% of all natural product "chemical space", including related compounds not made by any organism. That's still a lot, but it's much more manageable.

Mukund Chorghade, president of THINQ Pharma in Mumbai, India, calls it a "profound idea". He who believes it would be a boon for drug discovery because it could provide vast numbers of test compounds for developing new treatments. There are skeptics, however. Larry Overman, a synthetic organic chemist at the University of California, Irvine, calls Burke a "visionary", but points out that natural product molecules are far more complex in structure than biopolymers such as DNA and proteins. Overman wonders if automated synthesizers could reproduce such complexity.

It is not at all clear where the funding for Burke's "moonshot" would come from, either. It would be a hard sell in the best of times, and the present US political environment is unfavorable. The Trump Administration has a notable indifference, even loud contempt, for science; under the circumstances, funding for science will be under strain to stay level. Burke may have to wait for the political winds to change before he can hope to make any progress.

[TUE 20 MAR 18] US V MICROSOFT

* US V MICROSOFT: As discussed by an article from THEVERGE.com ("What's At Stake In The Microsoft Supreme Court Case" by Sarah Jeong, 26 February 2018), the rise of the internet has led to the emergence of thorny legal issues associated with a global data system. Consider, for example, this scenario: Alice lives in the USA, signs up for a free email account from a UK-based service, with the emails stored in a Canadian data center. Suppose the police in any of these three countries decide they want to get their hands on Alice's emails. Where do they go, and who do they ask?

This is the problem posed in the case of US V MICROSOFT, which went before the US Supreme Court at the end of February. The back story is that US law enforcement handed Microsoft a warrant -- a "Section 2703" warrant, defined under the "Stored Communications Act (SCA)" -- to get the emails of someone who had obtained an email account from Microsoft.

Microsoft gave the authorities metadata, such as the target's address book, but replied that the emails themselves were stored in a data center in Dublin, Ireland. Microsoft's position was that US law enforcement would then have to work through Irish authorities to get the emails. As Microsoft understands, the USA has mutual legal assistance treaties (MLATs) with about half the countries in the world, including Ireland, and so there is a legal mechanism in place. The Department of Justice wasn't happy with that answer. Microsoft had full access to Ireland-hosted contents, and Microsoft is a US company. The SCA covers access to domestic data, Microsoft is a domestic company, and so the DOJ felt the SCA warrant was entirely proper.

The case, in itself, seems picky, but there's issues of broad principle involved, which is of course why the case has gone to SCOTUS. Microsoft officials are not rejecting the principle of warranted access to customer data, instead being worried about the cross-border legitimacy of warrants, and getting caught in an international bind. Suppose, as a wild but not completely outrageous example, a Russian dissident has data stored in a server run by a Russian online firm, sited in the USA. Would Vladimir Putin then be able to get the data, bypassing American authorities? If so, that would make life very difficult for effectively all firms operating online, and make their clients very unhappy.

The DOJ sees the situation as simpler than that: an American company is obligated to provide data in reply to a legitimate warrant from the US government, and where the data is stored is irrelevant. It appears the DOJ is really concerned about precedent, and the emails are not such a big deal: the case is not new, it's been working its way up to SCOTUS for five years, but the US hasn't bothered to exercise the MLAT with Ireland to get the emails.

There's another layer of complexity in the issue. The SCA was passed as part of the Electronic Communications Privacy Act (ECPA) in 1986, before the concept "cloud computing" was even known. What happens, the DOJ asks, if data for a target is stored in a cloud storage network, with parts of the data stored in a half-dozen countries? Would the US have to deal with six different countries to get the data? What if the US doesn't have an MLAT with one of those countries?

Since there's nothing publicly known about the target or the emails, it's impossible to see if the DOJ's warrant for the emails is focusing on a mountain or a molehill. As far as SCOTUS is concerned, the bottom line being the propriety of warrants issued under the SCA in an era of global cloud computing. Microsoft's bottom line is a big step beyond that, with the company seeking a SCOTUS decision that drives legislation to extend the SCA into the 21st century -- and that, in itself, will only be a step towards a global legal framework that is going to take a lot of effort to construct. In the meantime, Microsoft isn't handing over the emails.

[ED: Congress passed the "Clarifying Lawful Overseas Use of Data Act (CLOUD)" Act in late March, which clarified the SCA -- saying that companies can be handed warrants for access to data on foreign servers, but provided mechanisms to allow companies to challenge the warrant on the basis of privacy concerns, and also extended the use of MLATs by permitting access to foreign data on the basis of executive agreements with foreign countries. The US V. MICROSOFT suit was then dismissed by SCOTUS, both sides being agreeable to the matter.]

[MON 19 MAR 18] UNDERSTANDING AI (16)

* UNDERSTANDING AI (16): The ongoing fuss over the threat of artificial intelligence systems eventually giving humanity the boot tends to distract attention from more practical and honestly significant issues with AI. As discussed by an article from THEVERGE.com ("Artificial Intelligence Is Going To Supercharge Surveillance" by James Vincent, 23 January 2018), one of the biggest real worries is that AI will make it much easier to spy on everyone.

Consider the widespread use of surveillance cameras. The more surveillance cameras there are, the harder it is to watch them at all times -- and so it becomes useful to have an AI do the watching, keeping a log of what happens, and alerting a human to something out of the ordinary. What happens when surveillance cameras are as common as streetlights, or maybe even mounted on streetlights? And complemented by a fleet of camera-carrying police drones on street patrol? Each camera would have its own AI, with the AIs networked together to form an "all-seeing eye".

Crimes are going to be very hard to pull off when cameras are everywhere. That's reassuring -- but it's not so reassuring to think that effectively everyone will be under public observation, effectively all the time. The system would be able to track a car by make, paint job, and license plate. The cops could also upload a mugshot to the system so that a suspect could be tracked everywhere. On top of that, there's also the possibility of biases creeping into the system -- for example, marking a group of black teens as troublemakers who actually being good kids.

This is not the present reality, but it's not science fiction, either. Consider the startup company IC Realtime and the firm's flagship product, named Ella. It's a software product, an app and web platform that uses AI to sift through video feeds; IC Realtime calls it "Google for CCTV". Ella can understand natural-language queries with a high degree of comprehension, zeroing in on clips to find, for example, a tabby cat, or a school uniform, or an F-15 fighter jet. Ella also takes feedback from users to refine its searches. With machine learning and lots of experience, Ella is going to get smarter over time. According to Matt Sailor, boss of IC Realtime:

BEGIN QUOTE:

END QUOTE

Ella is hosted on Google Cloud; it can search a single video feed, or thousands of them. Users pay a monthly fee for Ella, starting at $7 USD, then scaling up with the number of feeds. IC Realtime is targeting both businesses, at any scale, and individual consumers. As Sailor points out, there are plenty of cameras that have a degree of smarts, but not enough to tell the difference between "a bird and a break-in. No AI. No deep learning."

That's less true all the time, however. Consider, for example, Boulder AI, which sells stand-alone AI cameras, not dependent on an internet connection. The company offers "vision as a service", with founder Darren Odom saying: "The applications are really all over the board. Our platform's sold to companies in banking, energy. We've even got an application where we're looking at pizzas, determining if they're the right size and shape."

For an example, Odom mentions a customer from Idaho who had built a dam and, to meet environmental regulations, had to monitor the number of fish making it up a fish (water) ladder: "They used to have a person sitting with a window into this fish ladder, ticking off how many trout went by. Then they moved to video and someone watching it." The customer contacted Boulder, which came up with an AI CCTV TV system to do the job: "We really nailed fish species identification using computer vision. We are now 100% at identifying trout in Idaho."

As sophisticated as the products of IC Realtime and Boulder AI are, in a decade's time, they'll seem toylike. Basic AI research is ticking along rapidly; computing power continues to grow; while training datasets are also expanding at a fast clip, with information technology firms aggressively bulking them up. Two of the biggest datasets for video analysis have been generated by Facebook and YouTube. YouTube's dataset, for example, features more than 450,000 hours of labeled video that it hopes will spur "innovation and advancement in video understanding."

Both Facebook and YouTube need the ability to analyze video streams -- not merely to improve user search capabilities, but also to moderate content. Russian manipulation of the 2016 US elections was a turning point for the two firms, putting them under pressure to detect, identify, and neutralize the Black Hats, while ensuring they don't step on their honest users.

IC Realtime is trying to keep up with changing technology, currently working on new features like facial recognition; a longer-term ambition is to interpret events in a video stream. Sailor says he's spoken to potential clients in education, who want video systems that can spot signs of trouble among students -- for example pupils clumping together, as if about to start a fight.

Boulder AI is also working on the next generation, for example a system to analyze the behavior of people in a bank, Odom saying: "We're specifically looking for bad guys, and detecting the difference between a normal actor and someone acting out of bounds." The company has been using old security-camera footage for training -- but since such videos tend to be low quality, they've also been shooting their own training video, using actors. While Odom is necessarily not too specific about how the training videos are put together, he did say the obvious, that they provide facial expressions and actions useful to a smart security camera: "Our actors are doing things like crouching, pushing, over-the-shoulder glances."

What is unsettling in this is its inevitability. Schools, for their own safety and that of the students, need a pervasive smart surveillance capability; and for obvious reasons, so do banks. As far as fleets of police drones patrolling the streets go, there's no more a basis for legal objection to that than there is to police cruisers patrolling the streets. The difference is, the drones keep a closer watch. Cameras on every streetlight? If people object to that, then shouldn't they object to streetlights, too? In any case, there's no invasion of privacy in observing what happens in public places. [TO BE CONTINUED]

[FRI 16 MAR 18] AMERICA'S CONSTITUTION (6)

* AMERICA'S CONSTITUTION (6): Section 4 of Article I of the Constitution left the electoral procedures for Congressmen up to the states, subject to Congressional alterations; and specified that Congress must meet at least once a year:

BEGIN QUOTE:

END QUOTE

Section 5 outlined general internal rules and procedures for the Congress:

BEGIN QUOTE:

END QUOTE

Section 6 specified that members of Congress would be paid by the Federal government; that, by the "Speech or Debate Clause", they would have a degree of legal immunity; and that, by the "Ineligibility Clause" AKA "Sinecure Clause", they would be independent of the executive branch:

BEGIN QUOTE:

END QUOTE

Section 7 defined how Congress, through the "Origination Clause" AKA "Revenue Clause" generates a law; with the law then, as per the "Presentment Clause", handed to the president; and the president then, as per the "Order, Resolution, & Votes Clause" either signing or vetoing it, the veto only being overridden by a two-thirds vote of the entire Congress:

BEGIN QUOTE:

END QUOTE

[TO BE CONTINUED]

[THU 15 MAR 18] GIMMICKS & GADGETS

* GIMMICKS & GADGETS: As discussed by an article from WIRED.com ("Skype's Rolling Out End-to-End Encryption For Hundreds of Millions of People" by Lily Hay Newman, 11 January 2018), early in the year Microsoft announced that the company's Skype voice-IP service would now offer end-to-end encryption for audio calls, text, and multimedia messages, through a feature named "Private Conversations". Skype is using the open-source "Signal Protocol (SP)" to implement encryption, ensuring that only the devices sending and receiving in the conversation have unencrypted access to them. Servers can have no access to the contents of the message.

Early on, Skype was regarded as a secure service, since it incorporated strong encryption and a decentralized peer-to-peer network. However, after Microsoft bought up Skype, users noticed subtle changes in Skype's operation; the most privacy-conscious users began to avoid it, worrying that it might be aiding third-party and government wiretap surveillance.

Private Conversations is intended to reassure users that their conversations are protected. At the outset, Skype doesn't support video chat; and of course, though it protects the contents of communications, it doesn't hide "metadata", such as when a conversation takes place, and how long it lasts. According to Eva Galperin, director of cybersecurity at the digital rights group the Electronic Frontier Foundation:

BEGIN QUOTE:

END QUOTE

Since a provider has to know the metadata for its own operations, it can't be truly kept secret. The question of how much access the law has to the metadata is one the law is going to define, not the providers. Skype had been left behind in the race to give users secure communications, but is now trying to catch up. The relatively high public profile of Skype means it can have a lot of influence; but it also places a burden of proof on Skype to show that its secure communications really are secure.

* As discussed by an article from SCIENCEMAG.org ("Computer Chip Mimics Human Brain, With Light Beams For Neurons" by Matthew Hutson, 20 June 2017), "artificial neural networks (ANN)" are on the leading edge of computing, being the subject of intensive research. Work has focused on ANN chips with electronic analogues to neurons, and on streamlined digital processors that can readily emulate neurons in software. However, there is a third option, an "optical neural network (ONN)" -- in which light beams are used for neural computations.

An ONN is attractive because light beams are fast and can be processed in parallel by optical elements; ONNs also tend to be energy-efficient. ONNs are not a new idea, but in the past they've mostly been lab setups, consisting of tabletops full of precision mirrors and lenses, nothing that resembles practical systems. Now, researchers at the Massachusetts Institute of Technology (MIT) in Cambridge have managed to build an ONN in the form of a microchip.

The MIT chip is fabricated from silicon; it simulates a network of 16 neurons in four "layers" of four. Data enters the chip in the form of a laser beam split into four smaller beams. The brightness of each entering beam encodes input data, while the brightness of each exiting beam represents output data, a "solution".

While in transit through the chip, the paths of light cross and interact in ways that can amplify or weaken their individual intensities. These interference effects crossings simulate the way a signal from one neuron to another in the brain can be intensified or dampened based on the strength of the connection. The beams also pass through simulated neurons that further adjust their intensities.

The researchers then tested their optical neural network on a real-world task: recognizing vowel sounds. When trained on recordings of 90 people making four vowel sounds, a traditional computer simulating a 16-neuron network was right 92% of the time. The ONN was right 77% of the time -- but the researchers believe that it can do as well as or better than a computer simulation, while it's also faster and more efficient.

Alex Tait, an electrical engineer at Princeton University who is working on similar technology, says:

BEGIN QUOTE:

END QUOTE

The study's two primary authors, Yichen Shen, a physicist, and Nicholas Harris, an electrical engineer, both of MIT, believe their technology could have applications in data centers, autonomous cars, and national security, providing ANNs orders of magnitude faster than existing designs, while drawing orders of magnitude less power. The two are starting up a company and want to have technology on the market by the end of the decade.

* There was a buzz going around a few weeks ago about Amazon.com's Alexa smart speaker generating spontaneous fits of laughing. One report from Twitter described it as unnerving: "Lying in bed about to fall asleep when Alexa on my Amazon Echo Dot lets out a very loud and creepy laugh ... there's a good chance I get murdered tonight." Another commented: "So Alexa decided to laugh randomly while I was in the kitchen. Freaked us out. I thought a kid was laughing behind me."

While this sounds like pranksters at work, it was more a joke Amazon was playing on itself. After investigating, Amazon engineers found out Alexa was suffering from false positives, thinking it had got the command: "Alexa, laugh." The reports of users being baffled by the laughing suggests they were exciteable and not inclined to do a bit of research to figure out what was going on.

In any case, Amazon's engineers changed the command to something more distinctive and recognizable: "Alexa, can you laugh?" -- with Alexa responding: "Sure, I can laugh." -- and then laughing. Yes, there's always another bug in the system.

[WED 14 MAR 18] US NUCLEAR RENAISSANCE

* US NUCLEAR RENAISSANCE: As discussed by an article from AVIATIONWEEK.com ("The Pentagon Wants Its Nuclear Tomahawks Back", 17 January 2018), the Obama Administration had begun an effort to update America's nuclear arsenal. As revealed in a draft of the 2018 Nuclear Posture Review leaked by the HUFFINGTON POST, the Trump Administration is ratcheting the effort another step up, significant elements of the new plan being reviving production of plutonium cores for new warheads and reintroducing a sea-launched cruise missile.

The plan states that America will respond to the growing might of the nuclear forces of China and Russia, as well as emerging threats from North Korea, by modernizing its elderly nuclear arsenal of Cold War-era bombers, submarines, missiles and nuclear-qualified tactical fighters. The plan intends to accelerate modernization programs set in motion by the previous administration, including the:

• COLUMBIA-class ballistic-missile ("boomer") submarine.
  
  
• Northrop Grumman B-21 flying-wing bomber.
  
  
• The "Ground-Based Strategic Deterrence" missile, to update the ICBM force.
  
  
• Long-Range Stand-Off cruise missile.
  
  
• Nuclear-capable Lockheed Martin F-35A Joint Strike Fighter,
  
  
• B61-12 guided free-fall bomb -- this last item discussed here in 2016, as an element of the Obama Administration's nuclear modernization program.

The effort would also reverse Obama-era arms reduction initiatives by restoring the Navy's nuclear cruise missile capability, and retaining the air-delivered B83-1, the US's last megaton nuclear bomb. To counter Russia's "significant advantage" in nonstrategic nuclear weaponry and enhance the range of military options against China and North Korea, the Pentagon will also retrofit "a small number" of submarine-launched ballistic missiles (SLBM) with a low-yield nuclear warhead, and obtain a modern sea-launched cruise missile.

The new policy leaves open the door for the Navy to obtain more than the dozen COLUMBIA-class submarines originally planned. The Pentagon also wants to start an effort in 2020 to find a replacement for the Trident D5 fleet ballistic missile, which is now being life-extended for service on the OHIO- and COLUMBIA-class boomers.

Additional boomers and a new missile for them are long-term considerations, clearly beyond the horizon of the current administration. Over the shorter run, the Trump Administration plans to order the Department of Energy's National Nuclear Security Administration to scale up production of new plutonium pits -- the "active ingredient" of a Bomb -- to "at least" 80 per year by 2030, compared to zero today. Most of America's plutonium pits were produced in 1978-89, having been refurbished for additional years of service.

The Congressional Budget Office existing plans to sustain and modernize all three components of the nuclear triad will cost about $1.2 trillion USD over the next 30 years: $800 billion USD for operations and sustainment and $400 billion USD for modernization projects. The Trump administration's push to accelerate existing plans, while adding a low-yield nuclear warhead and modern sea-launched cruise missile will substantially increase that sum.

The Trump Administration's nuclear-weapons policy is an extension, not a reversal, of that of the Obama Administration, a response to the efforts of China and Russia to update their nuclear arsenals. The idea is to tell adversaries that they can't intimidate the US; to that end, the Trump Administration is not going to adopt a "no first use" policy, which had been considered by the previous administration.

The USA, in reality, is never going to fire off a nuke first. The entire arms buildup is entirely for psychological effect. The Trump Administration's push for "more usable" nukes doesn't indicate plans to use nukes, just to give the US more cards to play in the global nuclear chessgame. Nonetheless, the nuclear arms race was always absurd, with enormous sums spent on weapons that nobody dared use. A fair case can be made that, in the face of nuclear weapons buildups elsewhere, the US is compelled to respond; but in doing so, it appears that the only sensible long-term goal -- to eventually put the Bomb on the shelf -- has been mislaid.

[TUE 13 MAR 18] ENERGY MARKETS & BLOCKCHAIN

* ENERGY MARKETS & BLOCKCHAIN: As discussed by an article from REUTERS.com ("As Energy Markets Evolve, Blockchain Powers Up" by Jeremy Wagstaff, 22 December 2017), while the bitcoin crytocurrency is now going into convulsions, the "blockchain" scheme underlying bitcoin -- a "distributed ledger" of a series of cryptographically-protected transactions -- is catching on in a range of applications.

For example, energy startup companies have been using blockchain to support electricity sharing in microgrid trials from Texas to Tasmania for a year or so. Now they're moving on to commercial projects, leveraging blockchain for payments and trading on a city-wide or even national scale. According to Stephen Callahan, Vice President, Energy, Environment & Utilities, Global Strategy, at IBM: "What the internet did for communications, blockchain will do for trusted transactions, and the energy and utilities industry is no exception."

Blockchain solves two important problems in the online world: ensuring that transactions can be performed without participation of a trusted intermediary -- no need for lawyers -- and, on the other side of that same coin, ensuring that those transactions can't be tinkered with later. The energy sector has entered an era of change and diversification, with inevitable fragmentation and chaos. Blockchain is a tool that can handle the increasingly elaborate and decentralized transactions between users, large- and small-scale producers, retailers, traders, and utilities.

Along with the ability to keep a robust account of transactions, blockchain's use of "tokens" -- allocations of resources certified by a blockchain -- also offers a way to reward users for saving energy; and provides support for small-scale transactions between individual users with solar panels who are both producers and consumers, known as "prosumers". The ability to add "smart contracts" onto a blockchain also makes it possible for actions to generate automatic transactions down to the lowest level, where meters and computers could autonomously reconcile supply and demand. According to Daniel Sieck of US law firm Pepper Hamilton: "The prospect of being able to track particular electrons via a blockchain as they move onto or off the energy grid has captured the imagination of many companies."

The World Energy Council predicts that "decentralized / distributed" energy will grow from 5% of the market today to 25% in 2025. Although some big players like Shell, BP, and IBM have dabbled in blockchain, they're not all that serious about it for the present, being labeled "blockchain tourists". The real action in the present is with startup companies:

• Power Ledger of Australia raised $34 million AD ($26 million USD) in an "initial coin offering (ICO)" in October 2017 -- obtaining funds for commercial operation of microgrids in Thailand and India, plus two commercial buildings in West Australia. Power Ledger also recently launched a 200-customer trial microgrid with power retailer Origin Energy in Sydney.
  
  
• Energi Mine of the UK has created a blockchain-based platform to reward energy-saving users with tokens they can use to pay their energy bills or charge their electric vehicles. Energi Mine officials say the firm is already making money, if from the artificial intelligence side of its business.
  
  
• Electrify of Singapore has been operating a price comparison marketplace as the country liberalizes its electricity market. Electrify plans to launch a blockchain-based exchange for all consumers and producers next year, and is in talks with one of Japan's biggest utilities about doing something similar there.
  
  
• Grid+ of the US is going to launch its first retail scheme in 2018, using Ethereum blockchain technology to allow users -- traditional consumers, or renewables enthusiasts -- to buy and sell electricity at wholesale prices.

Energy startups raised about $200 million USD from ICOs in 2017, with a dozen more ICOs planned for 2018. Obstacles remain, including entrenched incumbents, and public worries about blockchain. To be sure, suspicions of blockchain are nowhere near as strong as those against the dodgy cryptocurrencies -- but it's still a largely unproven technology, and there's a bit of taint by association. Tony Masella, an energy consultant at Accenture, believes that blockchain-driven energy markets will emerge in a gradual evolution. Like the network protocols in the early days of the internet, blockchain "will drive dramatic transformation of the energy industry and unlock value" for everyone involved. "But it will not do so overnight."

[MON 12 MAR 18] UNDERSTANDING AI (15)

* UNDERSTANDING AI (15): An article on the rise of multicore processors was run here in 2011 -- providing hints that multicore processors with diverse cores were likely to become important for emerging artificial intelligence systems. As discussed by an article from ENGADGET.com ("What Do Made-For-AI Processors Really Do?" by Cherlynn Low, 15 December 2017), processors with AI capabilities are now increasingly common -- with Apple, Qualcomm, and Huawei having developed mobile chipsets with built-in facilities to support application machine-learning abilities.

Huawei launched its "Kirin 970" in 2017, saying it was the first chipset with a dedicated "neural processing unit (NPU)". Apple followed with the "A11 Bionic" chip, which drives the iPhone 8, 8 Plus, and X, with Apple claiming the chip features a neural engine that is "purpose-built for machine-learning." Late in the year, Qualcomm announced the Snapdragon 845, which partitions AI tasks to the most suitable cores.

These chips are all based on the concept of "heterogeneous computing" -- originally meaning a chip with different kinds of processors, using one processor for performance when it's needed, another for low power when performance isn't needed, and so sparing battery life. Contemporary smartphones often use the ARM big.LITTLE architecture, which takes exactly that approach.

The Kirin 970 and A11 Bionic take this concept a step further, by assigning machine-learning and such to dedicated neural units, which can execute such tasks rapidly, with relatively little power consumption. On the Kirin 970, the NPU takes handles tasks like scanning and translating words in pictures taken with Microsoft's Translator, the first third-party app to have been optimized for this chipset. Apple's A11 Bionic uses a neural engine in its GPU to speed up Face ID, Animoji, and some third-party apps.

The Snapdragon 845 doesn't actually have a dedicated neural unit; it appears to be a typical contemporary heterogeneous multicore chip with optimized software tools, and possibly some minor hardware tweaks. The Snapdragon 845 can use its digital signal processor (DSP) core to handle running tasks that require a lot of repetitive math, like listening out for a verbal cue; activities like image-recognition are better managed by the GPU.

It's difficult to see, particularly in the absence of well-established benchmarks, if Qualcomm is any worse off than the competition, and it's not so easy to see if the neural cores of the competition are much more special than what the Snapdragon 845 has. In any case, the use of AI cores offloads the CPUs of devices, improving performance. The AI cores also allow tasks that might have otherwise been farmed out to cloud to be run on the phone, which also enhances performance, as well as security.

Assignment of AI cores to specific tasks is not performed automatically by the operating system; applications developers use supported libraries like Google's TensorFlow, or more specifically its Lite mobile version, to determine which cores do what task. Qualcomm, Huawei, and Apple all work with the most popular software toolsets like TensorFlow Lite, and Facebook's Caffe2. Qualcomm also supports the newer "Open Neural Networks Exchange (ONNX)", while Apple provides a set of tools via the company's Core ML framework.

Multicore systems with AI cores are such a new technology that nobody has really done much of significance with them as of yet. However, nobody working in the field is skeptical of their value. AI means smarter and more convenient products, and the promise of innovative applications that are still to be invented.

* As discussed by a somewhat related article from TIME.com ("Your Future Smart Car Could Use AI To Help You Drive" by Lisa Eadicicco, 8 January 2018), at the 2018 Consumer Electronics Show, chipmaker Nvidia showed off two new AI software platforms for robocars. The first, "Drive IX", is intended to support the development of AI copilots that interpret both exterior and interior sensor inputs to assist the driver.

Drive IX, for example, could observe the driver with a camera, and take action if the driver is getting drowsy. The copilot could also provide an alarm if it spots a pedestrian, and the driver is looking someplace else. Drive IX will support gesture and voice-based control schemes as well.

The other platform, "Drive AR", will support augmented reality-based interfaces for cars that provide notifications and highlight points of interest. Nvidia's new Xavier processor will support Drive IX and Drive AR, with the company saying the Pegasus computer module based on the chip would be capable of Level 5 autonomy -- meaning a vehicle without a steering wheel, accelerator pedal, or mirrors. Of course, that's not saying Level 5 is going to happen in the near future. [TO BE CONTINUED]

[FRI 09 MAR 18] AMERICA'S CONSTITUTION (5)

* AMERICA'S CONSTITUTION (5): Article I of the Constitution established the Congress of the United States, with Section 1 -- the "Vesting Clause" -- declaring:

BEGIN QUOTE:

END QUOTE

Nine other sections followed; Sections 2 through 7 of Article I are most conveniently discussed together. Section 2 defined the structure of the House of Representatives:

BEGIN QUOTE:

END QUOTE

Section 3 similarly defined the structure of the Senate:

BEGIN QUOTE:

END QUOTE

[TO BE CONTINUED]

[THU 08 MAR 18] SCIENCE NOTES

* SCIENCE NOTES: As discussed by a release from the University of Texas Austin / UTA ("Why Poison Frogs Don't Poison Themselves", 21 September 2017), it's an obvious evolutionary gambit for prey species to become toxic, so nothing wants to eat them. However, evolution is blind, nothing is obvious to it, it simply goes in whatever directions it does.

Besides, becoming toxic poses a difficulty. For example, consider the little "poison dart frogs" of the jungles of Latin America. Each is loaded up with a mix of deadly neurotoxins, with the frogs being brightly colored to proclaim: YOU DON'T WANT TO EAT ME. The quandary is: since the frogs are so loaded up with toxins, then why don't they poison themselves?

UTA researchers probed this question by studying a subgroup of the poison frogs that use a neurotoxin named "epibatidine". It's a toxin that, when ingested by a predator along with a frog, binds to receptors in the predator's nervous system, jamming neural impulses, causing seizures and possibly death. The UTA researchers found that small mutation in the frogs -- a change in just three of the 2,500 amino acids that make up the neural receptor -- renders the toxin ineffective in a frog. In addition, precisely the same change appeared independently three times in the evolution of these frogs.

Working with partners in Ecuador, the UTA researchers collected tissue samples from 28 species of frogs -- including those that use epibatidine, those that use other toxins, and those that are not toxic. The researchers then sequenced the gene that encodes for the neural receptor for all 28 species, to construct an evolutionary tree for the gene.

The neural receptor normally accepts signaling molecules involved in cognitive processes such as learning and memory. Epibatidine mimics these molecules, binding to the receptor and strongly activating the neuron. It turned out that frogs that generate epibatine have a minor mutation that disrupts the binding of epibatidine, without rejecting the normal signaling molecules.

Neurotoxins can in principle be used as powerful painkillers, though it's proven difficult to handle the side effects. This study is a step towards a better understanding of how the neurotoxins work, and how they can be better put to good use. Incidentally, the relationship between epibatidine and the receptor mutation leads to the possibility that the mutation occurred first -- as a "neutral" mutation that did no harm and no good -- and then led to frogs that generated epibatidine with no harm to themselves.

* As discussed by an article from WIRED.com ("With Designer Bacteria, Crops Could One Day Fertilize Themselves" by Megan Molten, 14 September 2017), modern agriculture is supported by fertilizer production -- which is an energy-intensive process, and also leads to problems with run-off of fertilizer into watercourses. Farmers and consumers would be better off to minimize the use of fertilizer.

Legumes -- peanuts, peas, and many types of beans -- don't need so much fertilizer, because they host a group of microbes called "nitrogen fixers" that grow in the root hairs of their host plants, forming knobby nodes and converting free nitrogen in the soil to ammonia, the most significant component of fertilizer. Most of the world's biggest food crops -- corn, wheat, rice -- don't provide much or any support for nitrogen fixers.

In the era of genetic modification (GM), why should that be a problem? Couldn't GM microbes be developed that can do the job for any plant? Seeds could be coated with them before shipment to farmers. German biochem giant Bayer has now established a joint venture with Gingko Bioworks, a Boston-based synthetic biology startup, to do just that.

The joint venture will be run out of Ginkgo's new automated DNA foundry and Bayer Crop Science's R&D center in West Sacramento. Bayer's science team has already started screening its extensive microbial library for candidates to ship to Boston. The target is a microbe that can perform nitrogen fixing, but is also hardy -- able to stand up to production handling, and to survive without water for a long period of time, then activate when it gets wet.

To complicate matters, most nitrogen fixers don't culture well in a petri dish. It's one of the ironies of microbiology: pathogens tend to culture easily because they have to thrive in host organisms whose immune systems are trying to kill them off, while benign microbes need more agreeable environments. Worse, nitrogen fixation is tricky. At least 20 genes code for the proteins directly involved in turning free nitrogen into ammonia, and the process involves associated metabolic pathways. To make matters even more troublesome, not much is known about the interactions of nitrogen fixers with their immediate environments. What makes them thrive? What shuts them down?

Evolution works over deep time, with the relationship between legumes and nitrogen fixers honed over millennia. Can the scheme be tweaked to work with wheat, rice, or corn? Ginkgo co-founder and CEO Jason Kelly says: "That's going to be one of the core challenges of this. But what we have going for us is that the plant really wants this nitrogen, and historically that's the right scenario for symbiotic relationships arising. Evolution is on our side."

* As discussed by an article from SCIENCEMAG.org ("Surfing Fin-Embedded Sensors Collect Coastal Data", 8 September 2017) author Jon Cohen got to mix science with surfing by hitting the California waves -- with a surfboard featuring a sensor fin.

The fin was developed by engineer Phil Bresnahan and coastal biogeochemist Tyler Cyronak, both of the Scripps Institution of Oceanography in San Diego, and both surfers. With financial help from a New York City-based nonprofit named the Lost Bird Project, the two brewed up a surfboard fin crammed with a temperature sensor, a GPS receiver, a circuit board with a microcontroller, a Bluetooth chip, and a rechargeable battery. They hope to later add sensors pH, chlorophyll, salinity, and oxygen. For the moment, it's a pilot project, with 50 "smartfins" loaned out to surfers.

The intent of the exercise is to collect data on coastal zones, that data being a matter of interest for researchers who monitor the health of sea life-rich reefs, the mixing of atmospheric gases by breaking waves, riptides, pollutants, and the ocean's absorption of heat from global warming. Remote sensing from, say, satellites doesn't track natural activity in the near-coast regions very well. Surfers can do the job, and it's a surprisingly widespread practice. Even in the UK, not noted as a surfer's paradise, there are enough surfers to perform tens of millions of observations a year.

[WED 07 MAR 18] FIGHTING ROBOCALLS

* FIGHTING ROBOCALLS: As discussed by an article from WIRED.com ("The Robocall Nightmare Is Only Getting Worse -- But Help Is Here" by Lily Hay Newman, 20 November 2017), the plague of robocalls from sleazy telemarketers has reached a flood level, with billions of bogus calls afflicting Americans every month, tens of millions a day. It's been getting worse. The "Do Not Call" list has been around a long time, but it doesn't deter crooks -- the most it does is make it clear they're crooks who are blatantly flaunting the rules.

In November 2017, the US Federal Communications Commission (FCC) announced new rules that allow telephony providers to block some connections that appear to be robocalls before they get to end users. It's a welcome move, but not a complete solution. According to Jim McEachern, a senior technology consultant at the communication industry standards body ATIS: "It's a worthwhile thing to be doing, but it's not the be-all and end-all. The catch is that if you start blocking numbers that have [certain properties], then the robocallers will just spoof different numbers."

The FCC allowed telecom companies to block calls from:

• Invalid numbers, for example those with a fake area code.
  
  
• Numbers not linked to any service provider.
  
  
• Numbers not currently assigned.
  
  
• Numbers set up to make calls, but not receive them.

That should put a dent in robocalls for a while, but the scammers will adapt. McEachern says that the Canadian Radio-Television & Telecommunications Commission put in such rules some years ago; they helped, but hardly fixed the problem.

The FCC also acknowledged that the new rules could cause problems, notably blocking legitimate calls; the FCC recommended that providers come up with schemes for convenient reporting and resolution of such difficulties. Some of the FCC commissioners also were concerned with the provision allowing providers to charge for blocking. Verizon has been charging $3 USD per month for a robocall blocking service since June 2017, though other companies like T-Mobile and AT&T have so far offered blocking services for free.

In addition, fraudulent robocalls are only part, if a big part, of the problem, but end users also get unwanted calls from banks, debt collectors, and other financial institutions. There's been disagreement over whether such organizations have the right to call without consent, or continue to call when they've been told to cease and desist.

The FCC has been cracking down on robocallers on the basis of existing regulations. In August 2017, the agency fined Philip Roesel and his company Best Insurance Contracts INC more than $82 million USD for making a total of 21.5 million robocalls, in violation of the Truth In Caller ID Act. Roesel's company was making hundreds of thousands of calls a month, riding on unassigned phone numbers and spoofed caller ID information.

Along with the FCC, the Federal Trade Commission (FTC) is fighting robocalls, having initiated a new information-sharing initiative with telecoms in August 2017. It's nothing radical; the FTC merely sends consumer complaints about obnoxious phone calls to the telecoms to make sure everyone has the list of flagged phone numbers the FTC is constantly collecting.

More imaginatively, the FTC has also run a series of "Robocall Challenges," awarding thousands of dollars in prize money to projects like call-blocking apps. Popular anti-robocalling services like "RoboKiller" and "NomoRobo" both came out of the FTC challenges. Such apps often collect blacklists of robocalling numbers for blocking -- but many also have a mechanism to pre-answer suspicious calls before they ring through, assessing the voiceprint of the person in the recording, or the tricks the scam call is using to try to game the system, then enhancing their protection accordingly.

Block lists are a wall approach, a dimwit defense that just begs to be circumvented. ATIS has been working with the Internet Engineering Task Force, a standards body, on a smarter approach. The goal is to create interoperable standards that can be adopted by all mobile and VoIP calling services to do cryptographic digital call signing, so calls can be validated as originating from a legitimate source, and not a spoofed robocall system.

The scheme is built around "Secure Telephony Identity Revisited (STIR)" and "Secure Handling of Asserted information using toKENs (SHAKEN)" frameworks. They're based on digital certificates using public-key cryptography. STIR defines the construction and transmission of a digital signature for an outgoing call, while SHAKEN defines the security protocol using STIR.

The idea is that telephony service providers will have private and public encryption keys. On receiving an outgoing call, a provider will validate the call, then send a digital certificate, encrypted with the provider's private key. At the receiving end, SHAKEN uses the provider's public key and the certificate to validate the call. SHAKEN can't be faked by spooked calls; it's now in testing.

The main problem is that SHAKEN can't be used on landlines with older signaling protocols. It is a step in the right direction, however, and everyone realizes the problem will have to be addressed one step at a time. The will seems to be there. McEachern says: "It's in everyone's interest to figure this out. We're getting to the point where the consumer doesn't trust the telephone network, and that's bad for everyone."

ED: I long ago decided the phone network was a necessary evil. I turned off the ringer on my phone, and almost never pick up unless I'm expecting a call. When I do pick it up, most of the time it's spam. I find it absurd that the phone isn't just another facility on my PC, like email, where it's under much better control. I also don't like having to pay a surcharge for the phone; I don't for email, that's part of my ISP service, and I value email much more than the phone. I'm drifting towards a purely digital solution with Google Voice & Skype, but that may not be easy to achieve. I'll see.

[TUE 06 MAR 18] AMAZON DOES HEALTHCARE?

* AMAZON DOES HEALTHCARE? Online retail giant Amazon.com is moving into the US healthcare system, having recently decided to peddle pharmaceuticals -- to then link up with two other giants, holding company Berkshire Hathaway and financial firm Morgan Chase, in a joint venture to provide healthcare for their million-plus employees. As discussed by an article from HARVARD BUSINESS REVIEW ("What Could Amazon's Approach to Health Care Look Like?" by Robert S. Huckman , 6 February 2018), the entry of the aggressive and relentlessly innovative Amazon into the discouraging swamp of the US healthcare system offers possibilities for improvement.

Skeptics suggest that Amazon, a newcomer to healthcare, won't be able to budge the bloated and inert US medical sector. Optimists, of course, see no cause to be so pessimistic. The alliance between Amazon, Berkshire Hathaway, and Morgan Chase has deep pockets, and a lot of business expertise. More significantly, Amazon has a near-fanatical dedication to customer satisfaction -- not a mindset deeply in evidence in the US healthcare sector -- and Amazon customers are, by and large, indeed satisfied.

The company has focused on a number of successful strategies and tools to achieve that goal. First, Amazon has made transactions with customers seamless and reliable. In 2005, the company introduced "Amazon Prime", a scheme in which customers could get two-day delivery on products for a set yearly fee. Before then, customers had to juggle shipments to save costs; it was a pain for customers, it was a pain for Amazon. Prime, by establishing an "all you can eat" model, encouraged customer loyalty -- enhanced by offering perks for Prime subscribers, such as an "Amazon channel" video download service, providing a selection of video series and movies.

In 2007, Amazon similarly introduced "AmazonFresh", which provides same-day delivery of groceries in certain big cities. Amazon could bring the same convenience to healthcare, cutting through medical bureaucracy to make patients feel the system is serving them, not treating them like pigs at the trough. AmazonFresh raises the possibility of Amazon providing same-day delivery of prescriptions as well.

Amazon's acquisition of Whole Foods has also given the firm a "bricks & mortar" outlet chain that could offer basic health care services, like those provided by CVS MinuteClinic or Walgreens. Those precedents suggest that Amazon wouldn't be breaking any new ground in this respect, but Amazon would certainly be competitive, and likely to up the game. Primary health care is the least expensive part of the healthcare system, but has the biggest payoff.

Second, Amazon has mastered the art of passive data capture -- keeping close track of what its customers buy, while asking them for only the most essential data. Privacy advocates don't like the way Amazon tracks customers, but the customers usually don't mind; Amazon uses the data to give them better service, and knows better than to misuse the data. The company's new Amazon Go store in Seattle extends passive data capture into the offline world. Customers swipe their phones on entry, with cameras and other sensors keeping track of what they pick up; they just walk out the door, and are automatically billed.

Consider applying the same thinking to healthcare. A patient could simply walk into a clinic, swiping a phone on entry, with the visit carefully recorded and assessed; there's no need to write up a report, the system keeps track of everything, with the doctor and patient able to retrieve an automatically-generated report at the conclusion of the visit. In a hospital, tracking the movement of patients could show occupancy level and utilization of operating rooms -- data useful for long-term management, or for daily scheduling of staff work and patient appointments.

That leads to data analysis, the third element that Amazon can bring to healthcare. The company not only knows what each customer buys, it also knows how customers make their purchases: What purchases are difficult? Which ones are straightforward? If there's a difficulty, where's the bottleneck? Amazon will naturally be able to acquire comprehensive data on individual patients -- medical history, genome, relevant personal habits -- and just as naturally be able to pick out patterns from its customer base.

The fourth element follows the third in turn: the company's Zen ability to leverage work conducted for its own business into a customer business. The best-known example of this was the creation of Amazon's cloud computing unit, Amazon Web Services (AWS). At a 2003 strategic retreat at Amazon, company executives suggested that, having developed information technology (IT) systems for its own use, was presented with a opportunity to provide the same services, using cloud computing, to smaller and younger internet companies on a "pay-by-use" basis.

The scheme was attractive to internet startups because they not only didn't have to set up their own in-house IT systems, they didn't have to pay for anything more than they needed. AWS, while not Amazon's biggest money-maker, does a boom business. One of the big problems with US healthcare is the lack of standardization and interoperability of patient records; not only could Amazon impose standards on the industry, but could by the exact same coin also provide IT services that healthcare providers would find more attractive than trying to handle such by themselves.

What was particularly intriguing about Amazon's jump into cloud services was that, at the outset, the company did not have any more than a general notion of exactly what services customers needed. Amazon talked to the customers, the customers explained what they wanted, with Amazon then tailoring its services accordingly. Amazon is in much the same position relative to healthcare. The company has no real experience in the sector; all the company's executives have at present is a set of ideas to try.

That may end up being a prescription for failure; but it's at least as easy to think it may be a revolution, long past due, for America's dysfunctional healthcare system. How matters play out promises to be interesting.

[MON 05 MAR 18] UNDERSTANDING AI (14)

* UNDERSTANDING AI (14): Augmented reality appears to be another wide-open frontier for AI technology. Mobile apps like Snap, a messaging app, and the interactive game Pokemon Go pioneered widespread use of AR, but they only hinted at what might be made of it. AR's advocates see it as transformational, turning the internet into an experience accessed through a computer or smartphone display, into an integrated flow of data in the course of one's daily activities -- with AR devices offering capabilities such as real-time translation and facial recognition.

Big tech firms are only getting warmed up to AR. Google and Apple have launched AR software-development kits, hoping to encourage developers to build AR-infused apps to run on their platforms. Google got too far ahead of the learning curve with Glass, which flopped commercially; the company backtracked, but did not give up on the technology. Microsoft has developed a VR headset, named "HoloLens", but it is an expensive niche product. Facebook and Apple are believed to be planning offerings of their own. So far AR hasn't been a big winner, but nobody is discounting its long-term potential.

Another frontier that has big promise, but hasn't yet taken off, is autonomous vehicles. Tech firms are cruising the roads to build up proprietary road datasets to support robocars, while using computer vision to allow cars to recognize signs, other cars, and obstacles. Given that the global car market flies at an altitude of about $10 trillion USD, it's no wonder there's so much interest -- and AI developed for cars is likely to have a wide range of other applications, in drones and other robot systems.

Different companies are taking their own approaches. China's Baidu is working on a robocar operating system, following the model of Google's Android for mobile devices; it's unclear how Baidu plans to make money at it. Alphabet has its own robocar effort, as do Uber, Tesla, and a number of startups. Established carmakers, on the defensive, are investing as well, though seems Apple has scaled back its car ambitions.

The push into robocars is just another example of how tech firms see AI as an enabling technology across the board. The big tech firms are so committed to AI that Alphabet, Apple, Microsoft, and others are now developing their own "AI chips" -- instead of buying them from silicon vendors like NVidia, which currently owns the AI chip market, the company's hardware being used in a wide range of AI-oriented projects. In developing their own hardware, Alphabet, Apple, and the others are faced with the "standards dilemma": do they try to keep their hardware proprietary to keep competitors from getting an edge off their work, or do they attempt to establish standards that will boost the market as a whole?

An inclination to keep the hardware proprietary leads to the question of if big tech will grab more power by taking a lead in AI. They've got the mass, skills, and money to take a domineering lead -- Google having been particularly aware of that fact, getting in early, pioneering applications of AI technology, and obtaining a pool of talent. At the same time earlier revolutions, like the personal computer revolution of the 1980s, led to the ascent of new players, like Apple, that were more agile than and out-maneuvered the existing giants, like IBM.

There does seem to be an understanding among the giants that trying to create a walled garden in the AI world won't be to anyone's benefit over the longer run. Along with publishing papers, many companies today make their machine-learning software libraries open source, available to rivals and independent developers. Google's library, TensorFlow Lite, is particularly popular. Facebook has open-sourced two of its libraries, Caffe2 and Pytorch.

Open-sourcing has a benevolent aspect, but it's also enlightened self-interest, ensuring that the libraries get a lot of debugging. The publicity is all for the good as well. There are concerns that the owners of the libraries will start charging for them down the road; but there's no big concern over it. Everybody involved is so busy trying to stay on the leading edge in the present to worry too much about a future that's too open-ended right now to predict. That, of course, ends up being a worry in itself. [TO BE CONTINUED]

[FRI 02 MAR 18] AMERICA'S CONSTITUTION (4)

* AMERICA'S CONSTITUTION (4): The Constitution starts with a memorable preamble that eloquently distills the motives for its creation:

PREAMBLE:

END PREAMBLE

It wasn't just an exercise in eloquence; it flatly stated that the authority of the Constitution was derived from the citizens of all the United States. Although none of the states conducted referendums for approval, the people of the states voted for the delegates to the conventions, and a number of states relaxed their voting restrictions to increase the pool of voters. No state limited voting rights beyond the restrictions normal to that state.

While the Articles of Confederation had defined the central government as deriving its powers, such as they were, from the states, the Preamble of the Constitution made no mention of the states. State legislators were perfectly aware that they had been bypassed. In Virginia, Patrick Henry had zeroed in on the phrase "We the People" and the absence of reference to "states" to proclaim: "This government is not a Virginian, but an American government."

That was exactly correct, the Preamble having gone on from that to proclaim "a more perfect Union" -- not a confederacy of sovereign states, but truly a "United States", under the direction of a central government. The model in mind was the union of England and Scotland into Great Britain in 1707. Advocates of the new government, labeling themselves "Federalists", reassured doubters that the states would not simply disappear, instead becoming building blocks of an American union of states, the states retaining significant rights.

However, although the Federalists didn't emphasize it, they never tried to conceal or deny the fact that the Constitution included no provision allowing a state to withdraw from the United States. During the proceedings of the New York convention, skeptics of the new order had tried to install an escape hatch, seeking the right to secede if the new order proved unsatisfactory. After impassioned debate, with Alexander Hamilton leading the Federalist camp, the provision for secession was trounced -- the New York convention deciding to ratify on the assurance that the state's suggestions for amendments be fully considered.

Ratification was to be all-or-nothing; there could not be "a more perfect Union" if a state could arbitrarily leave in a snit. If a state wanted to leave, it would have to be with the assent of the Union. As reported in the papers, the attitude of Hamilton and John Jay, who was also involved in the New York convention, was that "a reservation of a right to withdraw ... was inconsistent with the Constitution, and was no ratification."

* In any case, the Preamble went on to establish the "mission statement" for the new United States, which was to "establish Justice, insure domestic Tranquility, provide for the common defense, promote the general Welfare, and secure the Blessings of Liberty to ourselves and our Posterity ... " This was only superficially like the mission statement of the Articles of Confederation, with the new government having a much less limited set of objectives:

• "To establish Justice": The implication was the establishment of just common laws across the United States, plus a functional and superior Federal judiciary to adjudicate those laws.
  
  
• "To ensure domestic tranquility": The Federal government could intervene in disputes between the states, with the states bound to respect the Federal judgement.
  
  
• "Provide for the common defense": The Federal government would maintain standing military forces to defend the United States, in particular to prevent interference by the European powers. States would not be burdened with the expense of standing forces, though they would retain militias.
  
  
• "Promote the general Welfare": The Federal government would have broad powers to override state actions that were contrary to the welfare of the collective -- in particular regulating interstate commerce to maintain a free trade environment -- as well as have powers to take actions beneficial to the whole.
  
  
• "Secure the Blessings of Liberty": The Federal government was to ensure the liberties of the American people, and not become an instrument of oppression.

It must be emphasized that, at the outset, the Constitution rested heavily on the concept of "States' Rights". While the Federal government called the shots on the actions for the United States as a collective, the ability of the central government to intervene in the internal affairs of the individual states was very limited. The states would never have agreed to the union had it not been so.

The Preamble, again, was a mission statement; it did not articulate any rights or responsibilities -- and was indeed not at all specific, being declarations of very broad principles open to interpretation. It has been invoked to this day in support of multiple principles in conflict with each other. It has never had much legal standing, though apparently it has been cited by the judiciary in rejecting ridiculously frivolous lawsuits.

The specific rights and responsibilities were outlined by the Articles that made up the rest of the Constitution, as well as the Amendments attached to it. It did, however, provide the underlying basis for the entirety of the Constitution, through assertion of its validity. For that reason, the Preamble is sometimes referred to as the "Establishment Clause".

The greatest significance of the Preamble lies in its first three words: "We the People". For all the grand reverence placed on the Framers in later times, the document did not begin with: "We the Framers". The men who created the Constitution saw themselves as agents of "the People", executing their will -- and though all those men had agendas, there is no reason to doubt they were sincere in their motives.

This distinction is not at all trivial, since it means that the question of: "What did the Framers intend?" -- is, if not irrelevant, less significant than the question of: "How does the Constitution serve the People?" Today, the Framers are all dead; but the collective of the people endures. [TO BE CONTINUED]

[THU 01 MAR 18] ANOTHER MONTH

* ANOTHER MONTH: As discussed by an article from THEVERGE.com ("Dreamscape Immersive's ALIEN ZOO Takes Guests To An Interstellar Jurassic Park" by Bryan Bishop, 23 February 2018), a Los Angeles startup named Dreamscape Immersive is now pushing forward to construct a public virtual reality entertainment system, the initial offering being titled ALIEN ZOO. It's now playing at the Westfield Mall in the city.

Visitors enjoy the experience in groups of up to six -- $20 USD a head in the pilot outlet, with tickets current sold out weeks in advance. After checking in, they're each given a head-mounted display, along with small trackers on hands and feet. Having kitted up, they enter the virtual world to stand on a platform that will take them on a 12-minute trip through an alien zoo on an orbital station, seemingly floating through an exotic landscape while they interact with, or defend themselves against, the inhabitants of the zoo.

ALIEN ZOO is the first product of the company, which was founded by former Disney Imagineering chief creative executive Bruce Vaughn, along with Hollywood producer and former DreamWorks president Walter Parkes. Their idea was to combine VR with physical sets, props, and sensations to give a holodeck-like experience. The wind blows on the visitors, while the platform rumbles as it "flies" through the air.

Dreamscape Immersive is not the only company working on such schemes, but the company is one of the most ambitious, seeing the possibility of outlets everywhere. While the past history of "digital theme parks" is not encouraging, it's simply not that expensive to set up an outlet, particularly as off-the-shelf VR gear gets cheaper. The active area is only about 18 square meters (200 square feet) in size, and the props to support the experience are not expensive -- a fan to produce wind, for example. Any bare-bones mall outlet can handle it.

Of course, it might be nice to have a TV observing the active area, so visitors can amuse themselves watching recordings of what they were actually doing during their visit to an imagined world. It doesn't require a permanent facility, any outlet with a big enough room will do the job.

Dreamscape Immersive officials want to expand the zoo, bringing in more zoo exhibits, meaning a visitor will only see a portion of them on any one visit. According to Parkes: "Think of it like the wild animal park, and right now 40 percent of it's built, and they're bringing people in -- but soon there's going to be the polar enclosure. The zoo is going to be that: build out other sections, so you could come back and take another route."

The company is planning more outlets, and is also working on other VR experiences. Again, Dreamscape Immersive may end up being just another novelty, to be discarded once the initial thrill wears off. Then again, it might become a feature of natural history museums and the like, as with IMAX theaters, with producers turning out new VR experiences on a regular basis. [ED: As of 2020, this appears to be going nowhere. Theaters have taken a brutal blow from the COVID-19 pandemic, and may not survive. In addition, VR causes motion sickness easily, and so does not have universal appeal.]

* Going from there, regarding adventures with my XBOX 360 / Kinect game machine last month, I finally got my copy of DISNEYLAND ADVENTURES. As mentioned, it provides a virtual Disneyland that players can explore, with "mini-games" -- 18 in all -- taking the place of the rides.

It took me a little while to get comfortable with the game. One of my biggest problems was figuring out how to exit properly, so I didn't have to simply turn off the XBOX -- to then be compelled to go through the time-consuming game setup again. I finally realized it was the standard Kinect alert: stand straight up, hold my left arm out at an angle, and a command screen will interrupt the game, allowing me to leave. Incidentally, I also hadn't realized that Kinect uses voice commands; I've been playing with them a little bit, but I need to understand the command system better.

The running around Disneyland is amusing, it's a fairly-well developed virtual world -- though it's troublesome to keep from bumping into people. Yes, they do get out of my way when I bump into them, but it still feels rude. However, the running around amounts to a game for kids, with them having encounters with Disney characters like Mickey Mouse and Donald Duck, collecting tokens as they do so. Although I may explore the grounds further, it's not much fun for adults.

I was after the mini-games anyway. I figured out that, after going through setup, I could hold both arms up and get a menu, one selection being a map that could take me to particular attractions and their associated mini-games. When I bought the game, I was primarily curious about the Space Mountain attraction -- which is, for those not familiar with it, an indoor roller-coaster that goes through dizzying starfields. It's a modest coaster, but the props are spectacular.

I figured the mini-game would be something like space surfing. It was no disappointment, though I actually got to ride a space scooter through warp gates, asteroid fields, and space battles. The only problem is that it's not very arduous; I need to check out some of the other mini-games to see if they give me a better workout. With 18 games, that should give me a lot to explore. It seems I can use voice commands to select a destination or a game; I'll have to give that a try.

Incidentally, the only avatars available in the game are, not surprisingly, kids. That made me uncomfortable; I'm an old bachelor, and us folk know to be nice to kids, but keep them at arm's distance -- for good reason, even an appearance of impropriety with kids can be very bad news. I found I was more uncomfortable with the boy avatar than the girl, so I cooked up a cute little blonde girl, dressed in black stretch pants and a light sweater. I found out that had a feminist angle, that it was something of a mind shift for an old guy to be taking the role of little girl.

I had been running the games off of the XBOX CD drive, but it took too long to load installments of the game. I plugged in an 8GB flash microdrive and installed the DISNEYLAND ADVENTURES game on it. It was a bit tricky to figure out how to install; the XBOX hand controller is the tool for the job, one simply highlights the game icon on the XBOX display, then presses the "X" button on the controller to get to a menu that includes an INSTALL selection. Game loading time got a lot faster. Incidentally, I still have to keep the game disk in the DVD drive, no doubt as an anti-piracy measure.

I decided to put the default KINECT ADVENTURES game on the flash drive, too, but there wasn't enough space on the flash drive. I figured I'd get a 16GB USB 3 flash drive; the XBOX ports are only USB 2, but I thought it would be best to make completely sure the flash drive wasn't any bottleneck in drive performance. However, I finally realized that I'd bought a set of 16GB USB 3 flash drives a few months back for no other reason than they were cheap, and seemed like a nice thing to have. I plugged one of them into the XBOX USB port, and installed both games on it. Hey, I figured they'd come in handy, and they did.

The XBOX 360 / Kinect is dated technology. It'll keep me happy for a while, but eventually I'll have to move up to something more interesting. Virtual reality games? I'll see what happens. Something to look forward to.

* February is, of course, a short month, which means I have to scramble to get things done. Late in the month, I looked in the cabinet under my bathroom sink, and found water seeping into it. After puzzling for a while, I finally realized that the toilet next to it was seeping into the floorboards.

Oh dear, I'd been going around with that toilet for over a quarter century -- some problem would crop up, and I'd have a challenge to figure out how to fix it, the fix not always being obvious. Now I was worried about flooring repairs, too, which was certain to be real pain.

Anyway, the first thing to do was get a new toilet -- so I went to Home Depot and bought a Kohler Cimmaron elevated toilet. Next problem was getting it installed, a problem aggravated by the fact that NE Colorado had just had a cold snap, and plumbers were out of the office, fixing busted pipes. I was having problems finding a plumber, and the ongoing seepage was only going to make matters worse the longer the problem went on.

I did get a plumber named Kyle from Ace Hi Plumbing in Loveland to come out on Friday afternoon. Along with the seepage, there was the problem that the base on which the old toilet was mounted was broken, and it wasn't possible to bolt the new toilet to the base. Kyle managed to brush my worries away: all he had to was mount a new "step-up" base on top of the old one, and then everything was right with the world. I was back up and running by the evening. The seepage didn't seem to have caused much damage, and would dry out.

I was thinking the whole fiasco would cost me over a thousand dollars, but it came in under $450 USD, so I was happy in the end. The Kohler toilet is functionally much the same as the old one -- it's a gravity flush toilet, an elderly technology -- but with small refinements. It has a more modern float mechanism, but I'd installed much the same in the old toilet in the course of repairs. It has a new "canister" valve mechanism in place of the troublesome old flapper valve, inclined to leaking and improper closure; I'll have to look it over to figure out how it works. Most significantly, it had a smaller reservoir tank, meaning less water per flush, the bowl being clearly designed to be more efficient.

So how did it get a strong flush with less water? As it turned out, the old bowl just had a round hole in the bottom, but the new toilet had a sluice channel -- which made flushing more straightforward, and in fact wouldn't work well with much more water. It didn't flush quite strongly enough at first, but I screwed the float up to ride higher in the tank, and also discovered that it's throttled, I hold down the handle for longer to get a stronger flush. Ah, the things that excite a home-owner! In any case, I'm glad to be rid of that troublesome old toilet.

* As for the Real Fake News for February, it began with an intense discussion in the Senate over raising the government debt limit, with Senate Democrats finally agreeing to a long-term deal that will keep the government running for two years. The Democrats were accused of "caving in" and not holding out for a deal on immigration -- but clearly Senate Majority Leader Chuck Schumer doesn't like the idea of using a government shutdown to get what he wants. Such extortion would reflect badly on the Democrats, alienating moderate voters. The moderate voters are the ones to worry about: the sorehead Left has no sensible alternative to voting Democrat.

The budget deal was played up as a victory for bipartisanship, but then the Republicans took a hard line on immigration, as did the White House -- with President Trump making threatening noises about deporting the "Dreamers", illegal immigrants brought to the USA as children and raised here. The tough rhetoric of GOP members of Congress may have been something of a show, setting the stakes high at the outset of the game. Yes, the Democrats are clearly in a disadvantaged position in the contest, but nobody really knows how the immigration debate is going to play out.

In the meantime, the investigation by special counsel Robert Mueller into Russian meddling in the 2016 US election continued, in the face of attempts by the White House and House Republicans to discredit it. They don't seem to be working, and were overshadowed when Mueller indicted 13 Russians on 16 February.

However, that event was overshadowed itself by a massacre at Marjory Stoneman Douglas High School in Parkland, Florida, on 14 February. The school was attacked by 19-year-old Nikolas Jakob Cruz, who went through the halls with an AR-15 assault rifle, shooting indiscriminately, with 17 killed and 14 wounded. At first, the national reaction was business as usual: outrage, followed by the usual declarations of useless principle by politicians, with no push for action.

The politicians did not factor in the kids from the school, who began a public campaign for tougher gun control laws, pointedly embarrassing politicians who echoed the National Rifle Association's (NRA) inflexible line against gun control. President Trump waffled on the issue, talking about gun control, then attacking the idea, and even proposing that high-school teachers be armed -- an idea which did not go over well with most high-school teachers.

The gun-control debate in Congress has, to no surprise, settled into yet another impasse, but the Parkland kids haven't shut up, with public outrage growing against the NRA. Right-wing trolls attempted to claim the kids were actually "paid actors", which raised hoots of derision, and led to the removal of a video by conspiracy-monger Alex Jones from YouTube. After years of going nowhere, the gun-control debate is now acquiring fearsome momentum. The issue will be hot in the upcoming mid-term elections.

Trump was also fumbling with embarrassments, such as an authenticated $130,000 USD payoff to a porn star named Stormy Daniels, to keep her from blabbing about an affair with Trump. One Karen McDougal, a Playboy playmate, came forward with a tale of another affair with Trump. Such things are by no means a surprise at this late date; indeed, we've got so used to them that most of us just shrug.

A survey of historians released during the month ranked Trump as America's most insignificant president, bumping up James Buchanan -- William Henry Harrison also ranked low, but that was because he only lived a month in office. Barack Obama jumped up to 8th place, no doubt continuing to look better by contrast with his successor. The poll was something of a shrug as well: it's plainly obvious that Trump, to the extent he will be remembered, will be seen in dead last place in presidential ranking by future generations. That doesn't concern either Trump or his fans, but over the long run, that doesn't matter either.

In the present, Trump's approval rating hovers in the range of 35% to 40%, his disapproval rate in the range of 55% to 60%. That's reached a steady state, the moving average being flat, and likely to remain so. His supporters aren't going to budge, and neither will his detractors. That is actually encouraging for his detractors, because an effective 3:2 ratio in their favor in a majoritarian society says who's going to be politically holding the cards.

Underlining the disorderliness of events, late in February the US Supreme Court refused to annul a lower-court ruling that blocked the Trump White House from deporting the Dreamers for the time being. It appears that SCOTUS simply wanted to make sure the White House would not be able to pressure Congress by holding the Dreamers hostage, giving Congress breathing room to come to an agreement on a matter in need of legislative clarification. That was followed on the last day of the month by the news that Jared Kushner, the president's son-in-law and one of his advisors, had been denied a security clearance -- a great embarrassment to the White House, if not to Trump, who knows no embarrassment.

The circus in the White House is a bit of a distraction. It's not so much that Trump is a bad leader, as he is the "un-president", no leader at all. The US has been decapitated, with a disorderly Right-wing mob haphazardly determining policy. The real issue all along has been the rise of extremism in the Republican Party, and the struggle against its pernicious effects on national policy. There being nothing much else for a bystander to do, the inclination is to tune it out. It is difficult, however, to think nothing of consequence is going on. Events have a momentum, it is just impossible to know where they are taking us.